Population pharmacokinetic model and limited sampling strategy for clozapine using plasma and dried blood spot samples

Abstract

Background: To improve efficacy, therapeutic drug monitoring is often used in clozapine therapy. Trough level monitoring is regular, but trough levels provide limited information about the pharmacokinetics of clozapine and exposure in time. The area under the concentration time curve (AUC) is generally valued as better marker of drug exposure in time but calculating AUC needs multiple sampling. An alternative approach is a limited sampling scheme in combination with a population pharmacokinetic model meant for Bayesian forecasting. Furthermore, multiple venepunctions can be a burden for the patient, whereas collecting samples by means of dried blood spot (DBS) sampling can facilitate AUC-monitoring, making it more patient friendly.

Objective: Development of a population pharmacokinetic model and limited sampling strategy for estimating AUC_0-12h (a twice-daily dosage regimen) and AUC_0-24h (a once-daily dosage regimen) of clozapine, using a combination of results from venepunctions and DBS sampling.

Method: From 15 schizophrenia patients, plasma and DBS samples were obtained before administration and 2, 4, 6, and 8 h after clozapine intake. MwPharm^® pharmacokinetic software was used to parameterize a population pharmacokinetic model and calculate limited sampling schemes.

Results: A three-point sampling strategy with samples at 2, 6, and 8 h after clozapine intake gave the best estimation of the clozapine AUC_0-12h and at 4, 10, and 11 h for the AUC_0-24h. For clinical practice, however, a two-point sampling strategy with sampling points at 2 and 6 h was sufficient to estimate AUC_0-12h and at 4 and 11 h for AUC_0-24h.

Conclusion: A pharmacokinetic model with a two-time point limited sampling strategy meant for Bayesian forecasting using DBS sampling gives a better prediction of the clozapine exposure in time, expressed as AUC, compared to trough level monitoring. This limited sampling strategy might therefore provide a more accurate prediction of effectiveness and occurrence of side effects compared to trough level monitoring. The use of DBS samples also makes the collection of clozapine samples easier and wider applicable.

Keywords: clozapine; dried blood spot; limited sampling strategy; pharmacokinetics; population pharmacokinetic model; therapeutic drug monitoring.

Figure 1.

Goodness-of-fit and weighted residual plots of the clozapine concentration: (a) population predicted clozapine levels based on the final model versus the actually measured clozapine levels, (b) individual predicted clozapine levels based on the final model versus the actually measured clozapine levels, (c) weighted residuals of the population predicted clozapine levels versus the predicted clozapine levels, (d) weighted residuals of the population predicted clozapine levels versus time. For modeling and simulation, six till 7 days of clozapine treatment was simulated to have steady state.

Figure 2.

AUC_(0-12h) based on the full curve compared to AUC_0-12h) calculated with the two-points sampling strategy with sampling at 2 and 6 h.