Caught in the current: Functional exploration of high-risk uveal melanoma biomarkers

This thesis aims to identify novel biomarkers in high-risk UM, and elucidating causality by extensive functional experiments; with focus on drug-able processes for clinical application.
Chapter 2 described the discovery of novel biomarkers in high-risk UM and AI-based prediction models. Based on transcriptomic data, a list of biomarkers correlated to poor-survival was generated for which the top hits are described in detail in Chapter 2.1. This list was the basis to study high-risk biomarkers in functional models described in Chapter 4. Chapter 2.2 describes an artificial intelligence box approach demonstrated there are morphological differences between low-risk and high-risk UM based on a simple histological stain. This black-box approach is thus able to identify morphological features to differentiate high-risk from low-risk; which could relate to biological processes.
However, correlation does not mean causation. In order to prove causation extensive experiments are needed with appropriate models and methods to test hypotheses. After discovery of novel biomarkers correlated to high-risk UM, these biomarkers need to undergo functional testing to prove causation. Next to in vitro models, animal models are the golden standard to validate causation. Despite extensive research and successful treatment of primary UM, there is a lack of therapeutic option for metastatic UM. Improvement of therapeutic options often relies on translatable animal models, which are sub-optimal in UM research. Chapter 3.1 summarizes all known mice and zebrafish models to gain insights where the field is lacking and how to improve UM animal model research. From this review, the rise of zebrafish xenografts was noteworthy and holds potential to be a quick and useful model for UM research. Chapter 3.2 describes the development of various zebrafish xenografts and automated methodology to provide a standardized read-out model for in vivo drug screening. Endogenous animal models are considered to hold higher translatability over xenograft-models; as these models are not immunocompromised. Chapter 3.3 describes a novel transgenic models that utilizes an promotor to drive genes of interest in a ocular specific fashion.
Once correlations are found and the appropriate methods are developed to investigate functionality; combining both research aspects together will give answers to proposed hypotheses. Identified biomarkers of interest found in Chapter 2 were studied by reverse genetics and causation was validated in vivo using models developed in Chapter 3. Top-ranking genes FOXD1 and RBFOX2, 2 transcription factors upregulated in high-risk UM, were studied in detail and their causal mechanisms are elucidated in Chapter 4.
All research techniques and findings have their strengths and weaknesses, some have direct implications, whereas others challenge our understanding. Discussing results is therefore paramount. Discussing results is therefore paramount. Chapter 5 discusses the findings of this thesis, its implications to progress UM research while; but also include their pitfalls.