Date Published: March 8, 2019
Publisher: Springer International Publishing
Author(s): E. A. M. Calvier, E. H. J. Krekels, T. N. Johnson, A. Rostami-Hodjegan, D. Tibboel, Catherijne A. J. Knibbe.
Previous research showed that scaling drug clearance from adults to children based on body weight alone is not accurate for all hepatically cleared drugs in very young children. This study systematically assesses the accuracy of scaling methods that, in addition to body weight, also take age-based variables into account for drugs undergoing hepatic metabolism in children younger than five years, namely scaling with (1) a body weight-based function using an age-dependent exponent (ADE) and (2) a body weight-based function with fixed exponent of 0.75 (AS0.75) combined with isoenzyme maturation functions (MFPBPK) similar to those implemented in physiologically based pharmacokinetic (PBPK) models (AS0.75 + MFPBPK). A PBPK-based simulation workflow was used, including hypothetical drugs with a wide range of properties and metabolized by different isoenzymes. Adult clearance values were scaled to seven typical children between one day and four years. Prediction errors of ± 50% were considered reasonably accurate. Isoenzyme maturation was found to be an important driver of changes in hepatic metabolic clearance in children younger than five years, which prevents the systematic accuracy of ADE scaling. AS0.75 + MFPBPK, when accounting for maturation of isoenzymes and microsomal protein per gram of liver (MPPGL), can reasonably accurately scale hepatic metabolic clearance for all low and intermediate extraction ratio drugs except for drugs binding to alpha-1-acid glycoprotein in neonates. As differences in the impact of changes in system-specific parameters on drugs with different properties yield differences in clearance ontogeny, it is unlikely that for the remaining drugs, scaling methods that do not take drug properties into account will be systematically accurate.
Accurate scaling of drug plasma clearance (CLp) from adults to children is important for the definition of first in child doses and hence robust study design involving younger children. To date, physiologically based pharmacokinetic (PBPK) models represent the most mechanistic method to scale CLp across the paediatric age range. PBPK models quantify the interactions between drug-specific and system-specific parameters and predict paediatric CLp by accounting for developmental changes in the system-specific parameters and how they impact drugs with specific properties. Application of these models is considered best practice in pharmaceutical industry, but obtaining PBPK ontogeny functions for a given drug is time-consuming and complex due to the requirement of a wide range of drug-specific and system-specific information. Moreover, all information may not always be available for each drug or each population. This leads to a need for simplified scaling functions which are more convenient for defining paediatric CLp in pharmacometrics. As multiple system-specific parameters may change in the paediatric population and as the impact of each of these changes on paediatric CLp may be different for each given drug with different characteristics, the challenge in developing simplified scaling functions is to aggregate all relevant information in functions with a limited number of scaling variables. Various simplified clearance scaling methods for the paediatric population have been proposed. Allometric scaling using a fixed exponent of 0.75 (AS0.75) is one of the simplest scaling methods, as it only uses body weight as scaling variable. However, AS0.75 has been shown to lead to large over-predictions of hepatic metabolic CLp in children younger than 5 years, especially when isoenzymes are immature (1,2).
A PBPK-based simulation workflow was used (5) that was running in R (a software environment for statistical computing and graphics) version 3.3.1 with R studio interface version 0.99.902 (6). In this workflow, ‘true’ adult and paediatric hepatic metabolic CLp values for hypothetical drugs with a wide range of properties that are substrates for known hepatic enzymes were generated using PBPK-based simulations, based on the dispersion model for hepatic metabolic CLp (7,8). This model was selected as it has been reported to better predict CLp than the well-stirred model for drugs with a high extraction ratio, while both models lead to equivalent CLp prediction for other drugs (7,9). Subsequently, the accuracy of scaling the ‘true’ adult CLp values to paediatric CLp values with the two scaling methods was assessed, by comparing CLp values scaled by ADE and AS0.75 + MFPBPK to ‘true’ paediatric CLp values.
As previous analyses have shown that hepatic metabolic CLp scaling based on body weight alone is not systematically accurate in patients younger than 5 years (2,16), the aim of this study was to systematically assess the hepatic metabolic CLp scaling accuracy of ADE and AS0.75 + MFPBPK in children younger than five years. Since this systematic assessment was performed using a PBPK-based simulation workflow analogue to previous analyses of other scaling methods (2,5,16), the reported accuracy of the different methods can be directly compared.