Title: Modeling exposure-response relationships in the rat self-administration model
Andrew C. Hooker (1), Elizabeth Dunn-Sims (3), David Fairman (2), Andy Mead (3), Piet Van Der Graaf (2) and Mats O. Karlsson (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden; (2) PDM Translational Research, Pfizer, Sandwich, UK; (3) Global Safety Pharmacology Pfizer, Sandwich, UK.
Background: There is a growing emphasis on the development of methodologies to derisk the abuse potential of drug candidates [1]. Numerous pre-clinical animal models have been proposed for investigation of the likelihood that a drug will sustain patterns of non-medical self-administration (abuse potential). One proposed model uses rats, with dual intravenous catheterization for infusion of drug and blood sampling, placed in a chamber with a lever that administers a possibly reinforcing test compound. Responding on the lever delivers a specific dose of the compound. The rats can then continue to respond on the lever to administer more of the compound; a specific time out between infusions limits the dosing. By varying dose per response the concentration at which reinforcing behavior occurs (if it occurs) can be determined. A limitation of the current paradigm is that it provides minimal insight into the exposure-response relationship and therefore can only be used to help define abuse potential in a qualitative manner.
Objective: To model the reinforcing behavior vs. concentration relationship and give a prediction of the dose regimen/exposure where no reinforcing behavior would be induced (with reasonable certainty) for an abused reference compound (cocaine).
Methods: Multiple study sessions with a total of 38 animals were performed. The studies included a single dose PK study (no lever presses) and numerous PKPD studies where dose per response was adjusted (including placebo) as well as the number of responses needed for each cocaine infusion. Studies were performed with animals at different stages of cocaine self-administration i.e.: those that have undergone repeated cocaine self-administration sessions (trained) and those in which the response has subsequently been extinguished (extinguished). A population PKPD model was then developed sequentially for this data using NONMEM.
Results: A two-compartment model described the PK data well. The PD variable time-to-infusion was modeled with a repeated time-to-event model. The rats exhibited different baseline characteristics if they were trained or extinguished (exhibit fewer level presses). Thus the placebo model included terms for both states. When cocaine was available, time-to-infusion decreased at very low concentrations, increased at mid-concentrations and at very high concentrations tended to decrease; a bell shaped hazard vs. concentration profile was used to describe this process.
Conclusions: A model that adequately describes the cocaine concentration-reinforcing behavior relationship for this animal model has been developed. With this model as a reference, new studies with other test compounds will be analyzed with the goal of creating a platform model to describe and predict patterns of non-medical self-administration for new compounds.
References:
[1]http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM198650.pdf
