What is PAGE?

We represent a community with a shared interest in data analysis using the population approach.


2001
   Basel, Switzerland

On mechanism based PK/PD modelling for direct drug-ligand binding systems

Philip Lowe Ph.D

Novartis Pharma, Drug Metabolism and Pharmacokinetics, CH-4002 Basel, Switzerland

The most common method to model drug effects normally involves either an empirical dose-response system, or a PK/PD system where the PD is dependant upon the PK. However, there are situations where a classical system does not work or is inappropriate, such as when there is direct and quantitatively significant binding of drug to a target ligand. Under these circumstances the kinetics of the endogenous ligand affect the drug PK. Two examples are given.

Desferal is the currently approved treatment for iron overload. The binding of desferrioxamine with iron to form ferrioxamine was modelled for individual thallasaemic patients based on a single dose slow-release versus standard formulation crossover study. Monte-Carlo predictions were then made to multiple dose scenarios, and compared with reality when the data were available. The predictions showed both the PK and urinary iron excretion (PD) to be non-linearly related to the dose of desferrioxamine. It was also predicted that the slow release formulation would not work at the doses tested. This was confirmed when the study data was obtained.

More recently, another compound with a similar mechanism of direct drug-ligand binding in the plasma has been studied. Instead of a two-stage process with individual WinNonlin fits followed by Monte-Carlo simulations based on geometric mean and SDLog parameters, NONMEM was used. All the single dose study patients were modelled simultaneously for 3 functions - PK, total and unbound ligand. Using this system, parameters were estimated for both the drug and endogenous ligand kinetics, together with the correlated variability terms. Using the simulation capabilities of NONMEM, different multiple dose regimen scenarios were explored, resulting in population predictions for the next clinical trial.



Top