Model-Informed approach for optimization of Pharmacokinetic studies design in the context of paediatric extrapolation: A step toward better decision making

Introduction: Extrapolation is a widely used concept in paediatric drug development, where information from one or more source populations is extended to a target population with limited knowledge. This often involves assessing the (favourable) benefit/risk balance (BRB) based on the similarity of pharmacokinetic (PK) exposures between adults and children. However, PK-based extrapolation requires generating PK data in children, necessitating appropriate clinical studies. Given the ethical and practical challenges in conducting such trials ensuring the relevance and informativeness of collected data is paramount. Regulatory agencies responsible for evaluating and approving extrapolation approaches face the challenge of ensuring their validity without imposing excessive demands, particularly given known constraints. This task is complex in the absence of well-established and objective approaches applicable to the majority of drugs.
Objectives: In this study, we employ a model-based approach to optimize key study design factors, including patient numbers, sample sizes, and sampling times, with the goal providing a framework for assessment of design key factors and informed decision-making, particularly in the context of paediatric PK studies conducted within extrapolation scenarios.
Methods: Three case-drugs were selected for this study: Meropenem, Mosunetuzumab, and Olanzapine, representing two small molecules and one monoclonal antibody. Both oral and parenteral administration routes were included. PK concentration data were simulated using published models and their associated parameters. Model parameters were re-estimated across 308 data collection scenarios and compared to the initial parameters. The simulations and parameter estimations were conducted using NONMEM version 7.3.0 coupled with PsN11, using the FOCE method with interaction. Stochastic simulation and estimation (SSE) cycles were executed on either the Cluster Hercules or a Windows computer. Subsequently, output folders were collected using Python Software, and the results were summarized using R.
Results: Our study shows how a simulation-based approach can optimize the design of paediatric PK studies for each of the three case-drugs. By using this approach, we can ensure that model parameter precision and accuracy are within acceptable thresholds, typically under 30% for precision and 20% for bias. For instance, to achieve the desired precision and bias for Meropenem clearance, our results suggest that the number of patients should range between 10 to 20, depending on the PK sampling scenarios.
Conclusion: The present research demonstrates how the design of a paediatric study can be optimized before data collection based on the available data from the adult drug development, that often precedes children’s. Our results show how, for each of the 3 case-drugs, using a simulation-based approach, paediatric PK study can be designed ensuring that model parameters precision and accuracy would be under predefined threshold, typically 30% and 20% respectively. These results provide useful information for drug sponsors and regulators as far as extrapolation in smaller populations such as paediatric is concerned.