Abstract
Allopurinol is a commonly used urate-lowering drug used for the long-term management of gout. Gout is an inflammatory arthritis caused by monosodium urate crystal deposition in joints. Reducing serum urate with allopurinol leads to the resolution of gout symptoms over time. The maintenance doses required to maintain a target serum urate of <0.36 mmol/L can range from 50-900mg daily and are difficult to predict. In addition, a slow dose escalation at therapy initiation means that it may take several months to reach an effective dose. A PKPD model would provide a means of predicting optimal maintenance doses facilitating the implementation of more cost-effective dose escalation strategies.
The relationship between allopurinol dose, the plasma concentrations of oxypurinol, the active metabolite of allopurinol, and urate-lowering response is unusual and remains unresolved despite several decades of clinical use. Equally, the influence of patient factors on allopurinol dose-response is often counter-intuitive and not well understood. These relationships are required to clarify maintenance dose requirements.
The overarching aim of this thesis was to clarify the factors that determine the dose-response of allopurinol and to use this information to inform allopurinol dose predictions.
The impact of genetic variability in transporters involved in the excretion of urate, including the T allele (Q141K) of ABCG2, on oxypurinol pharmacokinetics was explored in a population pharmacokinetic analysis. The findings demonstrated that transporter genotype does not influence oxypurinol pharmacokinetics.
The patient factors that determine the dose-exposure-response relationship of allopurinol were explored using a graphical and logistic regression analysis. Kidney function and diuretic use were the most important covariates that influenced this relationship. The odds of achieving the target urate were found to be predicted by kidney function, diuretic use, and baseline urate concentrations.
The implementation of novel methods for predicting kidney function and body size on oxypurinol population pharmacokinetics were conducted to create a universal model for oxypurinol across the human age range. The predictive performance of the novel oxypurinol pharmacokinetic model was evaluated against external data and was found to be similar to the model developed in Chapter 2.
A PKPD model was developed for allopurinol using a turnover model for urate. The best model was unstable and found to have imprecise and biased parameter estimates and inflated between subject variability. Predictions of urate-lowering response and allopurinol dose requirements were not conducted. Attempts to solve the model stability issues were explored leading to the hypothesis that the model may not be identifiable given the data available.
An identifiability analysis found that the PKPD model for allopurinol was not internally deterministically identifiable supporting the hypothesis that the values of oxypurinol elimination rate constant (ππ) and urate πππ’π‘ βflippedβ in some individuals creating an unstable model. The solution suggested was model simplification.
The utility of kinetic-pharmacodynamic (KPD) models in the setting of imperfect adherence was explored. The performance of PKPD and KPD models in describing the time course of drug effect was found to be similar during imperfect adherence.