Characterization of novel prognostic and histotype- specific biomarkers in ovarian cancer

Abstract

Epithelial ovarian cancer (EOC) remains the deadliest gynecological malignancy affecting women, and despite extensive efforts from the scientific community, tools for early detection have yet to fulfill statistical requirements for routine clinical use. Complicating matters further, the disease consists of several distinct histopathological subtypes with unique molecular and clinical characteristics. Thus, there is a considerable need for reliable molecular EOC biomarkers that can facilitate early detection, patient stratification, and improved prognostication. In this thesis, we comprehensively profile EOC histotypes at the transcriptional level in both early- and advanced-stage disease. Additionally, we present alternative analytical frameworks for survival analysis and epigenetic differentiation between groups. In the first study, transcriptional profiling of early-stage EOC samples was performed alongside predictive modeling using a proposed gene-panel, in combination with correlation analysis between DNA methylation and gene expression for proposed candidate markers. In the second study, epigenetic profiling of the same patient cohort was conducted using a novel approach for differential methylation analysis, together with predictive classification based on a proposed set of aberrantly methylated gene-promoter regions. The third study builds on study I by transcriptional profiling a cohort of advanced-stage EOC samples, identifying both diagnostic and prognostic histotype-specific biomarkers. In study IV, we use the cohorts from studies I, II, and III to enable a comparative analysis of the EOC histotypes in early- and advanced-stage disease. With the aim of identifying transcriptional signatures unique to tumor stage as well as markers shared across disease progression. Finally, in the fifth manuscript, relative survival analysis was applied to early- and advanced-stage samples to quantify years of life lost in genes with transcriptional states significantly associated with reduced expected survival. Taken together, this thesis presents a comprehensive analysis of the epigenetic and transcriptional landscape of EOC histotypes, identifying several promising candidate biomarkers that are reproducible across external datasets, and that hold potential clinical relevance for diagnosis, stratification, and prognostic assessment.