Preview

PDF, English Download (61MB) | Terms of use

Abstract

Pediatric cancer is a conglomerate of complicated diseases remaining one of the leading causes of death worldwide in patients aged 1 – 19. In recent years, major sequencing and precision oncology programs were launched and aggregated unprecedented amounts of data giving detailed insights into pediatric cancer at multiple omics levels. One possible avenue for improved treatment strategies, leverages synthetic lethality (SL) interactions between genes, potentially resulting in clinically relevant combination treatments. Despite the great interest, discovery of gene pairs with synthetically lethal interaction is very challenging and resource consuming because of the sheer size of the combinatorial space that needs to be covered. Advancements in in-silico prediction methods, utilizing multi-omics data, for SL interactions to narrow down the scope of investigation have gained popularity over the last years. In the first part of this study, I present and extensively evaluate a computational approach for the prediction of interacting pairs of genes exhibiting synthetic lethality. I apply my approach to two dedicated dataset I curated from multi-omics data of pediatric high-grade glioma (pedHGG) patients and compare the results. Finally, I describe a set of predicted SL pairs specific for pedHGG K27M, a particularly challenging childhood brain tumor, which includes multiple drug targets, targetable for example by HDAC inhibitors, that might serve as a guide for future investigations. Mutational signatures as a proxy for underlying mutational processes can be used as a biomarker for example for the detection of homologous recombination (HR) deficiency in adult cancer patients. However, biomarkers identified for adult cancer cannot be used for pediatric patients without further investigation and validation because of the different disease types and biological characteristics of pediatric cancer. In the second part of this study, I present an investigation into the active mutational processes in pediatric patients with disorders with abnormal DNA damage response. Using the latest set of mutational signatures and comparing two extraction algorithms, I thoroughly investigate associations of mutational signatures with clinically relevant characteristics and present a comprehensive overview of the mutational landscape in pediatric patients with abnormal DNA damage response. I was able to confirm reported associations of mutational signatures and my results further refine previous results especially with regard to differences between patients with germline TP53 mutation and patients with wildtype TP53. Further, I compare differences, reflected in mutational signatures, between patients with germline mutation in MSH6, MSH2, MLH1 or PMS2 that belong to the mismatch repair deficiency syndrome. Not just mutational signatures but also methylation can be used as a reliable biomarker e.g. to classify pediatric tumors with high accuracy into subgroups beyond what is possible via morphological differences. This classification method poises challenges for patients with hereditary abnormal DNA damage response and currently there is only limited knowledge about methylation patterns in such patients. In the third part of this study, I present new insights investigating methylation patterns in patients with abnormal DNA damage response. Using a molecularly characterized control cohort I assembled and accounting for immune cell infiltration, I was able to identify methylation signatures specific for defective DNA damage response across different tumor types. After detailed characterization of the discriminatory power of the identified methylation signature, that indicated achieving 90% precision is possible, I further investigated the biological function of identified methylation sites. This revealed association with biological functions including the RISC complex, RNAi and DNA damage response pathways such as base excision repair and nucleotide excision repair. Finally, I validated the presented methylation signatures in an additional internal and one external patient cohort consisting of liquid biopsy samples, demonstrating the broader applicability and highlighting a potential clinical application of the methylation signatures.