Ovarian cancer is one the most deadly malignancies in women with a five year survival of only 40%. Despite the fact that ovarian cancer, even the most common subtype of high grade serous ovarian cancer (HGSC), is a very heterogeneous disease most patients are treated with standard therapy consisting of cytoreductive surgery and platinum-based chemotherapy. Molecular characteristics are usually not considered. In this PhD thesis I evaluated potential molecular prognostic markers. Since in ovarian cancer peritoneal tumor dissemination is more frequent and important than distant metastases, the focus of this thesis was the characterization of two macroscopically distinct types of peritoneal spread, miliary (i.e. many small seeds) versus non-miliary (i.e. no/few larger implants), which we introduced in a publication arising from this thesis.
In the first study of this thesis comprising 200 epithelial ovarian cancer patients, tumor infiltration of cytotoxic T cells (CD8) was associated with prolonged overall survival; this effect seemed more important in patients with residual tumor. Patients with very low tumor cell proliferation (less than 5% of Ki67+ tumor cells) had a shorter overall survival, presumably due to worse response to chemotherapy.
In another project of this thesis we performed a comprehensive analysis of tumor cells from different peritoneal and ovarian sites and ascites from 25 HGSC patients and determined differences between the miliary and non-miliary tumor spread. Integrative analyses of RNA-sequencing of large and small (<200 nucleotides) RNAs, gene sets, pathways, and cell markers revealed a comprehensive picture of the two spread types. The non-miliary type has several classical features involved in metastasis and tumor progression such as epithelial to mesenchymal transition or higher levels of pro-angiogenic factors. Furthermore, non-miliary spreading tumors seem to provoke a more active immune system. In contrast, the miliary type is characterized by adaptation of several energy metabolism pathways and by a general downregulation of most coding genes (either caused by reduced transcription or by increased miRNA regulation). These tumor cells maintain their epithelial phenotype (e.g. more EpCAM+ ovarian tumor cells) and seem optimally adapted to peritoneal (in contrast to distant) metastasis. Furthermore, the miliary spread type was significantly correlated with worse survival as shown by using three independent patient cohorts and three different signatures (one from TCGA study and two developed by us to predict tumor spread: one based on large RNAs and one based on small RNAs). The small RNA signature comprises only 13 RNAs (RT-qPCR) and is applicable to formalin-fixed, paraffin-embedded tumors, which would render it a suitable method in clinical routine to determine the spread type of macroscopically undeterminable patients (with very advanced disease spread).
The revealed fundamentally different adaptation strategies of miliary and non-miliary could indicate distinct routes of dissemination. Further, differences in clinical outcome and molecular mechanisms and the worse outcome of patients with non-proliferating tumors underline the importance of specific therapies (e.g. according to tumor spread or proliferation).