Liver disease is a major cause of premature mortality, the incidence of which continues to rise steadily in the UK. Cirrhosis is the most common form, with a significant risk to patients who can deteriorate from acute decompensation (AD) to acute-on-chronic liver failure (ACLF). Mechanisms of cell death and immune dysfunction drive disease pathogenesis and have been linked to metabolic perturbations associated with the progression of cirrhosis to ACLF and short-term mortality. In this thesis, metabolic phenotypes of plasma and serum from patients with AD, ACLF and acute liver failure (ALF) were characterised using both 1H NMR spectroscopy and ultra-performance liquid chromatography – mass spectrometry (UPLC-MS). Untargeted 1H NMR spectroscopy and targeted UPLC-MS analysis of amino compounds, confirmed perturbations of metabolic pathways related to energy metabolism and the urea cycle, which were common to ALF, AD and ACLF. In addition, targeted UPLC-MS analysis of amino compounds identified unique metabolic perturbations for ACLF and ALF, which included increased concentrations of alanine, asparagine, glutamine, glycine, histidine, lysine, proline, threonine and tyrosine in ALF together with a reduced concentration of serine in ACLF. Furthermore, the quantitative amino compound data also indicated a greater metabolic perturbation in ALF compared to ACLF.
Untargeted UPLC-MS lipid profiling also identified metabolic perturbations that were both common and discriminant for ACLF, AD and ALF. Lipids including lysophosphatidylcholines (LysoPCs), sphingomyelins (SMs) and triglycerides (TGs) were depleted in these groups relative to healthy controls (HCs), whilst lysophosphatidic acids (LPA) were elevated and phosphatidylcholines (PC) were both elevated and depleted. Furthermore, lipid profiles generated from AD and ACLF patients were superior to standard clinical scores including CLIF- SOFA and MELD for predicting 30-day mortality based on AUROC analysis, with SM and LysoPC depletion associated with 30-day mortality in AD and ACLF patients.
In conclusion, this thesis identifies discriminant metabolic phenotypes of ACLF vs ALF and AD vs ACLF for the first time. In addition, metabolic changes associated with mortality risk are also identified, which highlights the potential for metabolic phenotyping to drive personalised predictive models of outcome in the clinical setting.