M Kågedal(1), MO Karlsson(2) AC Hooker(2)
(1) AstraZeneca R&D Södertälje, Sweden (2) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: Receptor occupancy can be assessed by Positron Emission Tomography (PET). Since PET measurements are very expensive it is important to ensure that the study is optimized to provide sufficient information based on only a few experiments. Generally, doses targeting 50% occupancy is perceived as most informative. The objective of the present work was to investigate how the doses in a PET-occupancy study should be allocated in order to maximize the precision of the derived relationship between exposure and occupancy dependent on the distribution of specific uptake in the brain.
Methods: Based on each PET-measurement, the partition coefficient, brain/plasma (KPt) can be derived for each region. Displacement of the radioligand from the receptor by a drug competing for the same binding site will result in a reduced KPt. The relationship between KPt and plasma concentration can be estimated by a saturation function where the affinity (Ki) is estimated as a parameter in the model. The occupancy can then be derived from the exposure and the estimated value of Ki. The study design considered included 6 volunteers with three PET-measurements in each volunteer.
The following situations were considered. a) A reference region void of receptors is included in the analysis. b) Two regions with different receptor densities are included. c) A single region of interest is included. In addition different random effects models were investigated. The D-optimal design criterion was compared to Ds with Ki as the parameter of interest. The optimal dose allocation was estimated using the optimal design tool PopED [1].
Results: If a reference region exists the precision of the Ki is less sensitive to the selected doses and no high dose with full saturation is needed. If no reference region exists, doses that result in close to full saturation improve the precision of the Ki estimate. When only a single region is included, near full saturation measurements are essential. With proportional residual error the optimal doses are higher compared to an additive residual error. Ds-optimal designs include more doses near the Ki value as compared to D-optimal designs.
Conclusions: The optimal dose allocation depends on assumptions regarding nonspecific brain uptake, distribution of receptors in the brain and the residual error. Targeting an occupancy of 50% is not always most informative. Ds-optimal design provides a mean to focus on the parameter of interest in the model.
References:
[1] PopED (http://poped.sf.net)
Reference: PAGE 21 (2012) Abstr 2552 [www.page-meeting.org/?abstract=2552]
Poster: Study Design