Eva Germovsek (1), Charlotte IS Barker (1, 2), Joseph F Standing (1)
(1) Institute of Child Heath, University College London, London, UK; (2) Institute for Infection and Immunity, St George’s, University of London, London, UK
Objectives: Clearance changes with both body size (usually measured by weight) and age due to maturation of processes such as glomerular filtration and drug metabolising enzymes [1]. As George Box famously said: “All models are wrong, some are useful”. Hence different modellers use different methods for parameterising clearance-related covariates for age and size. Different parameterisation leads to difficulties in comparing estimates between studies of similar drugs, and limits the possibility to learn about clinical maturation and size-related clearance (CL) changes. Arguments in favour of one parameterisation or another are usually based on fitting a single model to a distinct dataset [2, 3, 4]. We aimed to compare different parameterisations of maturation and size fitted to the same dataset, hence providing a clear way to compare distinct methods.
Methods: We conducted a systematic literature review of PubMed (search updated in April 2015) for publications containing information about CL of intravenous gentamicin (mainly renally cleared) and midazolam (mainly hepatically cleared) in three different subpopulations (neonates/infants, children, adults). We then searched for models for size and maturation, through systematic literature review and by writing to email discussion groups. The identified models were then fitted to CL data. Visual comparison of the fit of the models was undertaken and the mean prediction errors (MPE) and root mean square errors (RMSE) were compared.
Results: We identified 44 and 38 clearance reports for gentamicin and midazolam CL, respectively, and 18 distinct CL covariate models. Visual examination showed that models with fixed allometric exponent and no function describing organ maturation overpredicted the neonatal CL for both drugs. The best fit according to the Akaike information criterion and the MPEs provided CL models where allometric weight scaling was combined with a sigmoidal maturation function or the allometric exponent changed with age in a sigmoidal fashion.
Conclusions: Most models that included size and age described the CL of both drugs well, and were able to capture the neonatal and adult CL. Thus, we suggest using a combination of allometric weight scaling with a sigmoid maturation function, as exemplified here by gentamicin and midazolam, two drugs representing different elimination routes.
References:
[1] Kearns et al., N Engl J Med 2003; 349:1157-67
[2] Ince et al., Clin Pharmacokinet 2013; 52:555–565
[3] Wang et al., Clin Drug Investig 2013; 33:523–534
[4] Robbie et al., Antimicrob Agents Chemother 2012; 56(9):4927-36
Reference: PAGE 24 (2015) Abstr 3635 [www.page-meeting.org/?abstract=3635]
Poster: Drug/Disease modeling - Paediatrics