|Abstract: ||This thesis comprises of studies which collectively describe the development and application of novel metabonomic approaches to the investigation of xenobiotic metabolism and toxicity. Both computational and UPLC-MS technologies are critically evaluated to facilitate the development of rapid, robust and high throughput analytical approaches for the analysis of compounds in biological matrices, particularly applicable in the area of drug discovery.
Pattern recognition statistical modelling approaches were employed on both a small scale metabolism study of diclofenac and a large toxicology of acetaminophen. This methodology highlighted large differences between the test groups in both studies; however these discriminations were only compound related in the latter study. In a study with low statistical power, the probability of biologically unrelated processes, overwhelming the statistical significance of the compound related effects are quite high. Application of statistical normalisation for the correction of batch effects was also performed in an attempt to emolliate the issue of limited continuous LC-MS instrumental capacity. Batch correction was attempted using variance stabilisation with a glog function prior to median fold change normalisation. However, this approach was unable to correct the data to the extent where it was suitable for combined downstream analysis.
A rapid gradient microbore UPLC-MS method is then described to provide a high-throughput solution for metabolic phenotyping of large sample collections. This method increases the flow rate, decreases the column diameter and the gradient elution time, reducing the length of the analysis of each individual while retaining the diagnostic capability of the experiment.
The combined insights gained through this method development process are then applied to two separate toxicity studies of 2-bromophenol, one of acute and one of chronic exposure. UPLC-MS analysis coupled with 1H NMR spectroscopy identified the major metabolites as the sulphate and glucuronide conjugates, and other changes in endogenous metabolites. However the lack of reactive intermediates or other toxic metabolites indicate that 2-bromophenol is not a nephrotoxin. While these studies failed to discover any compound related toxicity, they have exemplified the utility of the metabonomic approach in this area.