Development of an Interactive Tool to Explore Doses & Sample for Paediatric Trials

Aims: The aim of this work was to develop an interactive tool that allows users to (1) explore paediatric dose levels that provide a similar exposure as predicted for an adult reference population, and (2) calculate the sample size required for a pharmacokinetic (PK) study in paediatrics based on FDA guidance [1,2].

Methods: A virtual demographic database of subjects aged >1 month to <65 years was created based on the Continuous NHANES database [3] and the WHO weight-for-age tables for paediatrics aged <10 years [4]. To account for age-dependent maturation of elimination processes, maturation functions reported in the literature for common renal and hepatic elimination pathways were implemented in the tool [5-7]. Users are required to input a reference dose in adult subjects, together with typical estimates of clearance, volume of distribution and their variability. Users may also specify up to three age subgroups for the paediatric study, and select multiple elimination pathways.

Results: The typical exposure in adults and paediatrics is calculated based on the area under the plasma concentration-time curve (AUC) at steady state, approximated as: AUC=Dose/Clearance. Typical clearance values are simulated for the adult and paediatric populations based on demographics (age, sex, and bodyweight) randomly sampled from the virtual subject database, taking allometry and enzymatic and renal maturation into account. 500 subjects are randomly sampled in each age group. A bodyweight based paediatric dose range between 1-70 dose units/kg (e.g. mg/kg) with a step-size of 1 dose unit/kg is explored in the exposure simulations for the paediatrics, with paediatric doses capped at the adult reference dose. Using the simulated clearance values, AUCs are calculated at each dose level to approximate exposure in all age groups. For each paediatric age group, the bodyweight based dose level providing the best agreement with the exposure at the absolute reference dose in adults is determined. The percentage of paediatric subjects with typical exposures falling within the 95% prediction interval of the typical adult exposure is used as the decision criterion (Figure 1).

The sample size for a paediatric PK trial is determined based on the minimum number of paediatric subjects required to achieve “a 95% confidence interval within 60% and 140% of the geometric mean estimate of clearance and volume of distribution for the drug in each paediatric group with at least 80% power” [1]. Virtual paediatric PK trials with 5-30 subjects in each age group are generated at random by sampling paediatric subjects from the virtual database. Individual clearance and volume of distribution values are then simulated based on the sampled demographics (age, sex, and bodyweight) and the adult BSV estimates, taking allometry and, for clearance, maturation into account. By default, this process is repeated 500 times for each sample size, i.e. 500 trials are simulated. The power to fulfil the FDA requirement is explored for each sample size across all simulated trials, and the minimum number of subjects achieving at least 80% power in each age group is determined [2].

 Conclusion: An interactive tool was developed to facilitate paediatric drug development.

David McDougall

  • University of Queensland