An Vermeulen Ph.D., Vladimir Piotrovsky Ph.D.
Johnson & Johnson Pharmaceutical Research & Development
Introduction: A depot injectable formulation of Risperidone (RIS, Risperdal CONSTATM) for i.m. administration once every two weeks has been developed and is on the market in quite a few countries now. The formulation is an aqueous suspension of RIS encapsulated in microspheres, which are composed of a biodegradable copolymer of poly-(d,l-lactide-co-glycolide). Polymer erosion controlles the release of risperidone.
The phase 3 clinical development program for this formulation included two efficacy trials: one double-blind trial (USA-121) and a non-inferiority trial (INT-61). The efficacy was evaluated using the Positive and Negative Symptoms Scale or PANSS.
RIS is mainly metabolised by CYP2D6 to the active metabolite 9-hydroxy-risperidone. The sum of RIS and 9-hydroxy-risperidone constitutes the ‘active moiety’ (AM) and is responsible for the improvement in PANSS score.
Objectives: The objectives of the current population analysis were:
· to model the pharmacokinetics of RIS and the AM in subjects of different phenotypes (USA-121);
· to model the treatment efficacy (improvement in PANSS scores) in terms of steady-state AM plasma concentrations (USA-121);
· to check the robustness of the PK-efficacy model by simulating the outcome of the non-inferiority trial (INT-61).
Methods and results:
Two population PK models were developed using USA-121 data: one for RIS and the other for the AM. Since the activity of CYP2D6 is subject to genetic polymorphism, the population is divided in poor, intermediate and extensive metabolisers, and this was taken into account in the RIS model.The models included a one-compartment disposition submodel characterized by clearance and volume of distribution and three parallel absorption pathways: an immediate pathway describing the absorption of non-encapsulated risperidone, and fast and slow sustained-release pathways. At the first stage, individual estimates of RIS and AM clearances together with other model parameters were obtained using data-rich trials. These estimates were summarized, and typical values and IIV were derived and further used as priors to obtain individual posterior clearance estimates for patients of sparse-data trials via a Bayesian procedure.
Individual clearances were then used to calculate model-predicted steady-state concentrations of the AM in USA-121, which were fed into a dynamic (indirect-response) population PK-efficacy model. It was based on the idea that the rate of increase and decrease of response determines its actual value. The concentration-response model has the following parameters: the PANSS score at patient inclusion, the PANSS score to be ultimately achieved due to disease progression, placebo and risperidone treatment, the rate constant of re-establishing the balance after a change in condition and the time to drop out. The effect of the AM was implemented as an Emax model.
The final population PK-efficacy model was used consecutively to simulate the INT-61 trial.
Reference: PAGE 12 (2003) Abstr 426 [www.page-meeting.org/?abstract=426]
Poster: poster