Prediction of Morphine Clearance in the Paediatric Population

Abstract

The pharmacokinetics of morphine in paediatrics have been widely studied using different approaches and modelling techniques. In this review, we explore advantages and disadvantages of the different data analysis techniques that have been applied, with specific focus on the accuracy of morphine clearance predictions by reported paediatric pharmacokinetic models. Twenty paediatric studies reported a wide range in morphine clearance values using traditional, rather descriptive methods. Clearance values were expressed per kilogram bodyweight, while maturation in clearance was described by comparing mean clearance per kilogram bodyweight between age-stratified subgroups. Population modelling allows for the analysis of sparse data, thereby limiting the burden to individual patients. Using this technique, continuous maturation profiles can be obtained on the basis of either fixed allometric scaling or comprehensive covariate analysis. While the models based on fixed allometric scaling resulted in complex maturation functions, all three paediatric population models for morphine yielded quite similar clearance predictions. The largest difference in clearance predictions between these three population models occurred in the first months of life, particularly in preterm neonates. Morphine clearance predictions by a physiologically based pharmacokinetic model were based on many continuous equations describing changes in underlying physiological processes across the full paediatric age range, and resulted in similar clearance predictions as well. However, preterm neonates could not be integrated in this model. In conclusion, the value of paediatric pharmacokinetic models is mostly dependent on clearance predictions and population concentration predictions, rather than on the individual description of data. For most pharmacokinetic models, however, the assessment of model performance was very limited and the accuracy of morphine clearance predictions as well as population concentration predictions was confirmed by formal evaluation and validation procedures for only one model.