Preview

PDF, English Download (20MB) | Lizenz: Exploring the Potential of the Human Microbiome for Colorectal and Pancreatic Cancer Screening by Cevirgen, Ece underlies the terms of Creative Commons Attribution-NonCommercial-NoDerivatives 4.0

Abstract

Microbiome research continues to move from purely descriptive approaches towards translation of findings into clinical applications. Microbiomics is a very active research field and the microbiome is now firmly established as an important factor in human health and disease, although mechanistic understanding currently remains limited. In this dissertation, I present two distinct studies which aim to (i) develop a microbiome-based screening approach for Colorectal Cancer (CRC), and (ii) investigate the microbiome of multiple body sites in Pancreatic Ductal Adenocarcinoma in order to describe a key microbiome signature that can be used for screening purposes. First, I developed a qPCR-based method for the early detection of CRC named Microbial Abundance-based Stool Test (MAST), which, after optimization and selection, is based on the quantification of seven selected microbial species from fecal samples. MAST showed a high predictive power and accuracy (AUC = 0.88) in a first screening cohort (77 controls, 56 CRC) and was further validated on an independent cohort of 344 individuals (272 controls, 72 CRC). To assess the potential usage of MAST in clinical applications, I compared MAST with established CRC detection method gFOBT which targets occult blood presence in stool. The combination of gFOBT and MAST resulted in increased specificity in the validation cohort owing to the complementarity of the captured signature (i.e., microbiome species abundance vs. occult blood presence). In summary, my work establishes MAST as a promising tool towards the early detection of CRC, and as a significant step in taking microbiome-based diagnostics towards the clinic. In my second project, I investigated the microbiome of multiple body sites of 57 Pancreatic Ductal Adenocarcinoma (PDAC) patients, 50 healthy controls and 29 chronic 13 pancreatitis patients by using 16S rRNA gene sequencing, metagenomic sequencing and fluorescent in-situ hybridization with the aim of establishing characteristic microbiome signatures of pancreatic cancer. While the overall microbiome community composition was not shifted between cases and controls in either saliva, stool or pancreatic tissue, the stool microbiome carried strong PDAC-specific signatures, distinguishing cancer patients from controls with high accuracy (AUC = 0.84 based on 22 microbiome species). The salivary microbiome, in contrast, showed only a weak signal overall. I validated the specificity of the fecal-based prediction model on external cohorts (3468 metagenomes) including healthy controls and patients affected by diseases such as type 1 or type 2 diabetes, liver cirrhosis, inflammatory bowel disease, and CRC. Moreover, pancreatic tumor and adjacent normal tissue showed markedly different microbiome profiles. Lastly, the presence of stool-related species/genera in both tumor and healthy tissue were validated by sequencing and in-situ hybridization methods. In summary, I describe a novel screening test for CRC, and advance the understanding of the microbiome in Pancreatic Cancer across several body sides.