Comparative analysis of scaling methods for dose selection in paediatric indications
Massimo Cella (1), Frank L. Kloprogge(1), Meindert Danhof (1) and Oscar Della Pasqua (1,2)
(1) Division of Pharmacology, Leiden/Amsterdam Center for Drug Research, Leiden, The Netherlands; (2) Clinical Pharmacology, Modelling & Simulation, GlaxoSmithKline, Greenford, UK
Objectives: Dose selection for paediatric indications remains a major challenge in early clinical development. Despite understanding of the role of developmental growth on pharmacokinetics and pharmacodynamics, medical practice often assumes a direct, linear relationships between body size, physiological function and clinical response. This assumption may lead to inappropriate dose selection in children. In the current investigation, we compare the accuracy of PK predictions obtained by different modelling approaches to scale pharmacokinetics from adults to children. The objective is identify which methods are best suited for the assessment of PK parameters in small populations.
Methods: The antimalarial drug atovaquone (ATV) is used as paradigm compound for our analysis. Nine clinical trials from GlaxoSmithKline’s clinical database were used for pharmacokinetic modelling. Different doses were available. The PK of ATV was characterised by a one-compartment model with 1st order absorption and 1st order elimination. The effect of different covariates on PK parameters was tested. Using the PK parameter estimates simulations were performed to estimate the individual systemic exposure (AUC). Taking into account the weight groups defined in the label of ATV, AUC distributions were summarised as median, 25th and 75th percentiles. Model performance was compared using alternative scaling methods. This evaluation is meant to support prospective data analysis in paediatric drug development, when availability of PK data in children is limited: Allometric scaling and the use of priors from adult parameter distribution.
Results: No significant differences were found between the AUC distributions predicted by the allometric scaling and the use of priors from pharmacokinetics in adults.
Conclusions: Both approaches can predict the exposures of atovaquone across different ranges of body weight with high precision. Incorporation of parameter distributions from pharmacokinetics in adults as priors for the modelling of paediatric data provides an alternative, valid method to evaluate drug exposure in children. In contrast to allometric scaling, which assumes a predefined relationship between body weight, CL and V, the use of priors enables the assessment of the correlation between parameters and covariates in the absence of rich data.
