Author:

Abstract:

Most patients with epithelial ovarian cancer (EOC) are still diagnosed in advanced stage disease, needing extensive debulking surgery and (neo)adjuvant platinum-based chemotherapy. Although initial response rates are good, most patients progressively develop platinum resistance during subsequent lines. EOC is a very heterogeneous disease, with different subtypes with distinct behavior and individually different responses to chemotherapy, even within the same histology. However, this heterogeneity is not reflected in the treatment, where the different subtypes are still treated as a single disease entity. The challenge for further optimization of treatment is thus to overcome resistance to platinum-based chemotherapy and to offer a more tailored therapy to the different subtypes. Unfortunately, there are no molecular markers in clinical use to either distinguish between patients with better and worse prognosis or to predict individual chemosensitivity. The aim of this thesis was to identify novel and validate some previously investigated genomic alterations in EOC and to search for the genomic alterations behind the differential chemotherapy response. In chapter 3, a current perspective on the genomic alterations of EOC is given, reviewing the most important studies regarding somatic mutations, copy number alterations, microRNAs, gene expression and DNA methylation in EOC. In a first study (chapter 4.1), 262 primary EOCs of mixed histology were genotyped for more than 100 hotspot mutations in genes potentially acting as prognostic markers in EOC. The main finding was that somatic mutations are limited to the, so-called, Type I tumors (low-grade serous, mucinous, endometrioid and clear cell), which generally respond worse to standard platinum-based chemotherapy. Of these, 49% were KRAS or PIK3CA mutant (versus 2.9% in the Type II tumors). Each histological subtype seemed to be characterized by distinct somatic mutations. PIK3CA mutations were predominantly found in clear cell and endometrioid carcinomas whereas mucinous subtypes harbored significantly more KRAS mutations. Within the serous histotype, low-grade tumors were more frequently KRAS or BRAF mutated than high-grade tumors. None of the mutations had a prognostic or predictive role. We concluded that the EOC histotypes, referred to as Type I, should not be regarded as one group but as distinct disease entities as they all have different driver mutations. We suggest that these tumors should not be treated with conventional chemotherapy but with targeted therapies based on their driver mutations. Secondly, we assessed somatic copy number alterations in a discovery (n = 86) and validation cohort (n = 115) of EOCs using high-resolution SNP arrays (chapter 4.2). We identified 53 regions to be significantly overrepresented in EOC, of which 6 correlated with overall survival (OS), progression-free survival (PFS) or platinum-free interval (PFI) in the discovery cohort. In the validation cohort, amplifications of the chromosomal region 14q32.33 also correlated with OS. Amplifications of 14q32.33 were also assessed in a pooled analysis involving both cohorts and published SCNA data from the Cancer Genome Atlas (n = 227). In this pooled analysis of 428 tumors, 14q32.33 amplifications significantly reduced OS, PFS and PFI. This region contains AKT1 as a potential driver gene, a gene which previously has been implicated as a key mediator of platinum resistance in cancer cell lines and mouse models. Moreover, AKT1 mRNA expression correlated with the number of chromosomal copies of the 14q32.33 region. These findings suggest that patients carrying 14q32.33 amplifications should benefit from a different therapy in first-line setting and could be offered treatment with an AKT-inhibitor. In a third study (chapter 4.3), a cohort of 100 advanced stage (FIGO IIb – IV) serous EOCs was analyzed for 13 microRNAs, selected from the literature based on their association with ovarian cancer. We confirmed the prognostic role of the miR-200 and let-7 family. In addition, we sought to identify microRNAs associated with differential chemotherapy response and OS via an unbiased genome-wide miRNA screen (n = 38). Unfortunately, none of the 25 selected miRNAs from the discovery cohort, could be confirmed in the independent validation set of 62 EOCs. However, a trend was seen for miR-193b in relation to PFI, but our numbers were too small to draw firm conclusions. In the search for new targeted therapies to improve survival in EOC, the epidermal growth factor receptor (EGFR) has been considered a promising target. In a last part of this thesis, we evaluated in a randomized phase III study the efficacy of erlotinib, an EGFR tyrosine kinase inhibitor, as maintenance therapy after first-line chemotherapy (chapter 5.1). Unfortunately, this strategy did not improve PFS or OS. We searched for biomarkers to predict which patients would benefit from this treatment (chapter 5.2). No subgroup could be identified with improved effect of erlotinib, based on immunohistochemistry or FISH for EGFR, or mutations in genes related to the EGFR pathway. However, patients with a positive FISH EGFR status had a worse OS (46.1 months versus 67.0 months) and PFS (9.6 months versus 16.1 months) than those with a negative status, making it an attractive molecular biomarker for further investigation as a criterion for selecting patients for prospective studies on erlotinib or other anti-EGFR therapies in EOC. Overall, specific genomic alterations associated with differential therapy response and patient outcome were identified in this thesis. It is clear that individualized and targeted therapy, based on the molecular drivers in the tumor, is proposed to be the future treatment strategy of ovarian cancer patients. The discovery of predictive biomarkers that identify patients which benefit from these targeted therapies is paramount to the success of these treatments.