A systems toxicology framework for improving the identification of paracetamol overdose

Abstract

Paracetamol (APAP) overdose is the leading cause of acute liver failure and a concerning global health issue. However, the current clinical treatment framework is heavily criticized for its sub-optimality. Within this thesis, a systems toxicology approach is taken in an attempt to provide further insight into the APAP overdose problem, and propose potential improvements to the current treatment framework. In Chapter 2, a proof-of-concept pre-clinical pharmacokinetic-pharmacodynamic (PKPD) model describing APAP metabolism and corresponding toxicity biomarkers (ALT, HMGB1, full K18, fragmented K18) is defined. A statistical model is combined with the PKPD framework to simulate in silico population groups with the aim of predicting initial APAP dose, time since overdose, and probability of liver injury. In chapter 3, an identifiability analysis is performed on the PKPD model to identify parameter uncertainties. The model is also extended, enabling predictions for individuals deemed both “healthy” and “high-risk”. In 2017 I was awarded the in vitro toxicology society mini-fellowship award, which funded 4 weeks of training in experimental wet-lab techniques. The training was used to investigate the effects of various combinations of APAP and its antidote, N’Acetylcysteine (NAC), on in vitro hepatocyte functionality. Subsequently, in chapter 4, the effect of the antidote (NAC) is incorporated into the PKPD model structure, and an additional toxicity measure is defined, describing severe loss of cell functionality. Different NAC regimens are tested, investigating their effect on both of our proposed toxicity measures. Through collaboration with the Royal Infirmary, Edinburgh, we obtained access to a clinical dataset of approximately 3,600 APAP overdose patients. In Chapter 5, a population-pharmacokinetic (Pop-PK) APAP model is defined, with PK parameters optimised based on this dataset. The framework has the ability to account for random inter-individual differences in PK parameter values. Current clinical toxicity thresholds are investigated and compared to those proposed by our model for various demographic groups.