Spinu, N (2021) Modelling of quantitative Adverse Outcome Pathways. Doctoral thesis, Liverpool John Moores University.

Preview

Text 2021spinuphd.pdf - Published Version Available under License Creative Commons Attribution Non-commercial. Download (16MB) | Preview

Abstract

With the growth of green chemistry initiatives, there is a demand for improved regulatory assessment of human exposure to exogenous factors. Proposed a decade ago, the adverse outcome pathway (AOP) framework serves as a knowledge assembly, evaluation, interpretation, and communication tool, designed to support pathway-oriented chemical risk assessment (CRA). The increasing number of resources and advances in machine learning (ML), artificial intelligence (AI), and the quantification of AOPs (qAOPs) has allowed for the integration of a variety of data streams including new approach methodologies (NAMs). These may predict causally inferred tipping points of the relationships that characterise a disease/adverse effect across multiple levels of biological organisation. This thesis aimed to provide an in-depth analysis of the qAOP concept and reinforces the types of efforts required to achieve validation, harmonisation and regulatory acceptance of qAOP models. The first part of this thesis assesses available qAOP models against a series of predefined common features, which enabled the challenges and opportunities for improving current practices to be identified. The second part of this thesis proposes improved methodologies for qAOPs, including the derivation of a network of linear AOPs that better depicts the complexity of biological effects and quantification of a simplified mechanistic AOP network based on domain knowledge and topology analysis. The thesis ends with a case study focused on the identification of empirical quantitative data associated with a linear AOP for quantification purposes. To apply the methodologies formulated, neurotoxicity, represented by neurodegenerative diseases such as impairment of cognitive function and Parkinsonian motor deficits, was studied. Lastly, the role of causality and reasons of why pattern-recognition is not sufficient to translate qualitative/mechanistic information into predictive models are discussed. Overall, the findings contribute to the advancement of the qAOP framework by expanding the knowledge, proposing recommendations and setting future directions towards the development and regulatory and scientific consensus of causal predictive qAOP models in toxicology. Other benefits to the field of study include how to combine information from linear AOPs into a more realistic representation of biological processes for the development of predictive models and the identification of which information (from alternatives) would be required for toxicological understanding. The work underlines knowledge gaps that need to be addressed, and exemplifies how to make use of, and integrate, the variety of available evidence for more informed predictions and improved decision making.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	in silico toxicology; qAOP model; AOP network
Subjects:	R Medicine > RM Therapeutics. Pharmacology
Divisions:	Pharmacy & Biomolecular Sciences
Date Deposited:	18 May 2021 08:34
Last Modified:	30 Aug 2022 16:04
DOI or ID number:	10.24377/LJMU.t.00015012
Supervisors:	Cronin, M, Madden, J and Enoch, S
URI:	https://researchonline.ljmu.ac.uk/id/eprint/15012

View Item