Rik Schoemaker and Adam Cohen
Centre for Human Drug Research, Leiden, The Netherlands.
When trying to fit model equations to experimental data, the situation may arise that individual subjects do not provide sufficient information to obtain adequate parameter estimates. This may be due to the fact that not all aspects are exhibited by all subjects or that the models are simply too complex. A solution is presented by applying non-linear mixed effect modelling (using NONMEM) to the data, which sensibly integrates the information provided by different subjects. The methodology will be illustrated by three different examples: -Determination of two-compartment pharmacokinetics of a novel drug in a rising dose design, where the lower doses provide insufficient information (due to assay limitations) to estimate the terminal part of the curve. -Determination of the kinetics of the low molecular weight heparin Clexane after IV administration using anti-Xa activity, effectively dealing with lingering low/basal activity. -Simultaneous estimation of the kinetics of the low molecular weight heparin Fragmin after subcutaneous and intravenous administration using APTT response.
These examples illustrate that use of NONMEM in complex models and data rich situations (e.g. in phase I) provides a powerful addition to the clinical pharmacologist's toolbox.