Evaluation of morphine pharmacokinetic models in neonates and infants.
Elke H.J. Krekels(1,2), J.G. Coen van Hasselt (1), Dick Tibboel(2), Monique van Dijk(2), Sinno H.P. Simons(2), Caroline D. van der Marel(2), Richard A. van Lingen(3), Anne M. Lynn(4), Imti Choonara(5), Meindert Danhof(1), Catherijne A.J. Knibbe(1,2,6)
(1) Division of Pharmacology, Leiden/Amsterdam Center for Drug Research, Leiden, The Netherlands; (2) Department of Pediatric Surgery, Erasmus MC-Sophia Children’s Hospital, Rotterdam, The Netherlands; (3) Princess Amalia Department of Pediatrics, Division of Neonatology, Isala Clinics, Zwolle, The Netherlands; (4) Seattle Children’s Hospital, University of Washington, Seattle, WA, USA; (5) Academic Division of Child Health, Derbyshire Children’s Hospital / University of Nottingham, The Medical School, Derby, UK; (6) Department of Clinical Pharmacy, St. Antonius Hospital, Nieuwegein, The Netherlands
Objectives: There is an ongoing debate on how to incorporate the influence of developmental growth in paediatric pharmacokinetic (PK) models. Bodyweight can be included a priori as a covariate by the use of a bodyweight-based allometric equation with a fixed exponent of 0.75 for clearance and 1 for distribution volume, or a more systematic approach in which bodyweight is regarded a covariate as any other can be applied. The performance of two population PK models for morphine in (pre)term neonates and children up to 3 years, developed using these two approaches, is evaluated in this study.
Methods: Using the same datasets [1,2], two population PK models were developed for morphine in this young population. Models were either based on a systematic data-analysis as described in Knibbe et al. [3], or on allometric principles based on bodyweight with fixed exponential scaling factors as described in Bouwmeester et al.[4] and Anand et al.[5]. The models were validated internally and externally using up to 4 datasets [6-9].
Results: In the analysis prediction-based diagnostics (basic goodness-of-fit plots), simulation-based diagnostics (NPDE), numerical diagnostics (bootstrap) and residual-type diagnostics (MDAWR, MAWR etc.) were evaluated, as were number of parameters needed to describe the data, ε- and η-shrinkage and the ease with which dosing recommendations can be derived. Neither of the models proved to be superior on all performed diagnostics.
To compare individual and population predictions from each model, individual POSTHOC parameters and population parameter predictions were plotted against each other. These plots show that both models predict comparable clearance values and distribution volumes, except for individuals at the boundaries of the weight-range.
Conclusions: Both methodologies for population PK model development in the paediatric population yield models that describe morphine PK in children younger than 3 years adequately. The choice of methodology depends on the aim and objective of the study. The model with the fixed exponential scaling factors is possibly more useful for extrapolations to older age-groups, whereas the model that was developed using a systematic analysis has fewer parameters. Moreover dosing regimens can be more easily derived from the latter model.
References:
[1] Van Dijk et al. PAIN (2002) 98(3); 305-313
[2] Simons et al. JAMA (2003) 298(18); 2419-2427
[3] Knibbe et al. Clin. Pharmacokinet. Accepted for publication
[4] Bouwmeester et al. Br. J. Anaesth. (2004) 92(2); 208-217
[5] Anand et al. Br. J. Anaesth. (2008) 101(5); 680:689
[6] Van der Marel et al. BJA (2007) 98(3); 372-379
[7] Van Lingen. Thesis "Pain assessment and analgesia in The newborn: an integrated approach"(2000)
[8] Lynn et al. Pain (2000) 88(1); 89-95
[9] Choonara et al. Br. J. Clin. Pharmacol (1992) 34; 434-437