Exploring the value of population pharmacokinetic and pharmacodynamic modelling: in infectious disease, cancer and diabetes.

Abstract

Dose selection is a critical part of individualising pharmacotherapy in order to ensure optimal safety and effectiveness. Population pharmacokinetic and/or pharmacodynamic modelling paired with Bayesian Forecasting is a sophisticated tool that can provide guidance for dose selection to clinicians. The overall aim of this thesis was to demonstrate the use of population modelling in three different settings and the utility of this approach for providing individualised dose selection.

In the first application the aim was to demonstrate that plasma samples of ribavirin in early Hepatitis C therapy could be used to predict plasma concentrations at Week 4 (critical for toxicity). Bayesian Forecasting was shown to provide a better prediction of ribavirin concentrations than standard linear regression analysis. This work demonstrated that safer and more effective doses could be predicted early in the course of therapy allowing earlier dose individualisation.

In the second application the aim was to confirm the published concentration-response relationship of imatinib in treating gastrointestinal stromal tumours. Interestingly, the data showed that patients with higher total plasma concentrations had a poorer response, opposite to what was expected. More work is needed to address this lack of concordance with the previous findings. One suggestion is that since imatinib is highly bound, unbound concentrations may be a better correlate of response than that based on total concentration.

In the final application, the aim was to establish a metformin concentration-weight loss relationship in a group of overweight, non-diabetic women. A small but significant relationship between total metformin exposure and weight loss was shown. This finding is novel, as a relationship between drug concentration and weight loss has not been presented in the literature previously.

Overall, the work presented in this thesis provided evidence for the value of population modelling and Bayesian Forecasting to assist in dose individualisation. Future directions for this work are: to refine the approach to dose selection in these and other important clinical areas of medicine; to establish a strong evidence base for implementation more widely; and then to pursue implementation strategies so that clinicians have access to these tools to make more appropriate dose selection.