Abstracts

Rationale: To scale an existing adult population pharmacokinetic/pharmacodynamic (PK/PD) model for brivaracetam (BRV) into children, using a combined adult-pediatric PK/PD model for levetiracetam (LEV), a compound with a similar primary mechanism of action, and to predict the effective dose of BRV in children aged 4 to 16 years.Methods: A population PK/PD model has been previously developed to describe the relationship between BRV plasma concentrations and seizure frequency change from baseline in adult subjects with partial-onset seizures (POS) in confirmatory efficacy studies. A pediatric population PK model is currently available for BRV. For LEV, both PK and PD data are available in adults and children. In order to support the extrapolation of PD in BRV to children aged ≥4 to <16 years, a LEV adult/pediatric PK/PD model was built and subsequently used to scale the existing BRV adult PK/PD model into children. The developed LEV adult/pediatric PK/PD model described seizure counts using a negative binomial distribution taking previous day seizure frequencies into account, and using a mixture model to separate a 'placebo like' and a 'responder' sub-population. Visual predictive checks (VPCs) were used to ascertain the ability of the LEV adult/pediatric PK/PD model to adequately simulate trial outcome in terms of percentage change in seizure frequency from baseline, and fraction of subjects with ≥50% decrease in seizure frequency. The validated LEV adult/pediatric PK/PD model was then used to support the scaling of the existing adult BRV PK/PD model into children. PK and PD simulations for BRV were performed in children for a range of mg/kg doses to predict BRV effect in pediatric subjects, and aid study design decisions.Results: The combined adult/pediatric LEV PK/PD model was able to describe both the adult and the pediatric data using the same drug effect population parameters, using a model structure very similar to the existing adult PK/PD BRV model. For LEV, 32.6% of subjects were estimated to be in the 'mixture model responders' sub-population. VPCs illustrated that the LEV adult/pediatric PK/PD model was capable of simulating the observed trial outcomes (Figure 1). Simulation with the adult BRV PK/PD model in combination with the pediatric BRV population PK model, allowed characterization of the pediatric dose-response curve, which predicts that the maximum response should be obtained with BRV 4 mg/kg/day dosing in children (Figure 2).Conclusions: The developed LEV adult pediatric PK/PD model showed that no scaling was needed for drug related PD parameters between adults and children. Therefore, the existing adult BRV population PK/PD model could be applied to the pediatric BRV population PK data, and allowed prediction of BRV drug effects in pediatric patients to support future studies. Maximum response is expected to occur around 4 mg/kg/day dosing of BRV in children aged ≥4 to <16 years, with a predicted -30.1% change in seizure frequency from placebo across this pediatric population. UCB supported.