Integrative multi-omic analytics for the early detection of pancreatic cancer
Citation:Kane, Laura, Integrative multi-omic analytics for the early detection of pancreatic cancer, Trinity College Dublin, School of Medicine, Surgery, 2023
Laura Kane_PhD Thesis_August 2023.pdf (Thesis) 10.77Mb
Pancreatic cancer (PC) has the lowest 5-year survival rate of any cancer, at just 12% in 2023. The poor survival rates observed in this cancer are the result of an accumulation of shortcomings and gaps in many aspects of PC research, with early detection and diagnosis being at the forefront of concerns. PC symptoms are notoriously vague and everyday in nature. Issues such as appetite loss, abdominal pain or weight loss are early warning signs, however, most patients ignore these common aliments, regarding them as unimportant. As such, the majority of patients diagnosed with PC are found at a late stage of cancer development, further contributing to the poor survival rates of this cancer. Moreover, the only FDA-approved biomarker for PC diagnosis, carbohydrate antigen 19-9 (CA19-9), is known to perform poorly in patients with co-morbidities such as diabetes or pancreatitis, making its utility in this setting extremely limited. Pancreatic cystic lesions (PCLs) are fluid-filled protrusions either on, or inside, the pancreas, which can be benign or pre-malignant. PCLs in many cases are precursors to PC, and as such could be the key to early detection of this cancer. Unfortunately, the current guidelines used to stratify patients based on their PC risk are imperfect. At present, there are several sets of these guidelines utilised globally, highlighting the lack of consensus among clinicians as to the best approach for these patients. PCLs represent a unique opportunity to identify and monitor patients with a high-risk of PC development, and as such more robust solutions to the stratification of these lesions are of urgent need. Here, the vast accumulation of research into potential biomarkers for PC is explored. A systematic review and meta-analysis of blood-based biomarkers highlights over 40 years of PC research, examining 250 manuscripts, and puts a spotlight on several promising biomarkers. Multi-biomarker panels are shown to be superior to single biomarkers alone, indicating that panels of multiple biomarkers more adequately adapt to patient-to-patient variability, allowing other biomarkers to make up the shortfall when one is dysregulated; a luxury not shared by single biomarkers. Moreover, it is shown that the current standard biomarker for PC diagnosis, CA19-9, has little utility on its own, but when added to a panel of multiple biomarkers, it can improve the panel. The study cohort utilised for biomarker identification was demonstrated to greatly affect the results obtained. Indeed, cohorts of both healthy individuals and patients with a benign condition were found to be more clinically relevant and produce more robust results than cohorts of either alone. Importantly, 13 novel biomarkers were identified as being repeatedly examined across the literature, with promising qualitative metrics for PC diagnosis. This systematic review of the literature also allowed the detailed critique of current PC research, with several common issues in study design and data reporting being highlighted across the included studies. Taking the information and evident flaws in PC research that were revealed via the systematic review and meta-analysis, this study then delved into biomarker identification and validation. Indeed, the performance of those promising diagnostic biomarkers identified in the systematic review was examined in a cohort of patient pancreatic cyst fluid (PCF) and blood sera, to determine their utility for risk stratification of patients with PCLs. While some were found to be differentially expressed in patients at a low- or high-risk of PC, their ability to stratify patients based on expression of these markers was poor. As such, exploratory multi-omic profiling of patient PCF and serum was conducted, focusing on the proteome and transcriptome of both. Strict differential expression analysis identified proteins and miRNA in both biofluids that were differentially expressed in low- and high-risk patients. Alone, these panels of proteins and miRNA perform poorly-to-moderately in stratifying patients based on their risk. However, when coupled to form multi-omic panels, these biomarkers perform risk stratification with high accuracy. Finally, integration of both the PCF-based multi-omic panel and the serum-based multi-omic panel was conducted using CombiROC software. This cross-biofluid multi-omic biomarker panel demonstrated the best stratification performance overall, and was uniquely capable of appropriately controlling for outlier patients with genetic mutations. While validation of these results in an independent patient cohort remains to be conducted, this promising biomarker panel could be the key to early detection of PC. Lastly, the biological activity of the PCF was interrogated. As PCLs can be precursors to PC, the role that the fluid within these cysts plays in the progression of PCLs to PC remains to be elucidated. Here, normal and intermediary pancreatic cell lines were exposed to patient PCF at low concentrations for 24 h. PCF was demonstrated to be biologically active, having the ability to alter cell line viability, proliferation, apoptosis, metabolism, phenotypic and functional marker expression levels, and DNA damage levels. Furthermore, PCF was shown to cause functional changes to the cells, resulting in increased invasive potential. However, PCF in many cases was also shown to be extremely cytotoxic to cells, even at low concentrations for just 24 h. Indeed, two distinct effects were seen in the PCF, cytotoxic and stimulatory. These preliminary results demonstrate the biological activity of PCF, and raise questions about its potential role in the progression of PCLs to PC. Further work is urgently needed to interrogate the mechanisms at play during the PCL to PC transformation, and the part that PCF plays in this process. If PCF could be demonstrated to contribute to the malignant transformation of PCLs, the aspiration of all PCLs as a preventative measure could be called for. This would have substantial implications for the management of PCL patients, and also for the advent of potential preventative measures for PC development. Overall, this body of work highlights the desperate need for better understanding of PCLs and PC, and presents promising avenues for future research in biomarker identification and early detection of PC.
Author: Kane, Laura
Publisher:Trinity College Dublin. School of Medicine. Discipline of Surgery
Type of material:Thesis
Availability:Full text available