External Validation of a Population Pharmacokinetic Model for Morphine and its Metabolites in Children under 3 Years who Underwent Cardiothoracic Surgery
Maria Xiberras (1,2), Gerdien Zeilmaker-Roest (2), Swantje Völler (1), Elke H.J. Krekels (1,3), A.J. (Bram) Valkenburg (4), Dick Tibboel (2), Enno Wildschut (2), Catherijne A.J. Knibbe (1,5)
(1)Division of Systems Pharmacology and Pharmacy, Leiden Academic Centre for Drug Research, Leiden University, Leiden, The Netherlands , (2) Intensive Care and Department of Paediatric Surgery, Erasmus University Medical Center–Sophia Children’s Hospital, Rotterdam, The Netherlands (3) Certara Inc, Princeton, NJ, USA, (4) Department of Anesthesiology and Intensive Care, Isala, Zwolle, the Netherlands, (5) Department of Clinical Pharmacy, St. Antonius Hospital, Nieuwegein, The Netherlands
Objectives: Morphine is widely used for postoperative pain management in children. Previously, a model was developed by Valkenburg et al. to describe the pharmacokinetics of morphine and its metabolites in children after cardiothoracic surgery (1). Compared to a model for children after major non-cardiac surgery, children after cardiac surgery showed a lower glucuronidation clearance of morphine alongside a lower elimination clearance of the metabolites, and higher estimated peripheral volume of distribution (1–3). The aim of the present study was to externally evaluate the population pharmacokinetic model for morphine and its metabolites in children after cardiothoracic surgery developed by Valkenburg et al. in a new independent dataset of children of similar age after cardiothoracic surgery and to extrapolate to wider ranges in age and weight.
Methods:
Data from the Paediatric Analgesia after Cardiac Surgery (PACS) study, which included patients under 3 years of age who underwent cardiac surgery with use of cardiopulmonary bypass (CPB) in 4 different centres in the BeNeLux area, were available for external validation. In the PACS study, the patients were randomized into two groups either receiving intravenous (IV) morphine or IV paracetamol as standard pain treatment. Patients from both groups could receive additional rescue morphine doses if numeric rating scale scores for pain exceeded predetermined cutoff values. From this PACS study, observed morphine concentrations in both groups were included in the current external model validation of the model by Valkenburg et al (1). NONMEM®VII (ICON Development Solutions, Ellicott City, MD, USA) was used for validation and model fitting with Bayesian re-estimation (i.e., MAXEVAL=0). The predictive performance of the models was evaluated based on a) basic goodness-of-fit plots for the overall population and stratified for relevant covariates, with specific focus on population predictions (PRED), b) normalized prediction distribution errors (NPDE), c) ETA distribution versus covariates.
Results:
A total of 699 samples analysed for morphine, morphine-3-glucuronide, and morphine-6-glucuronide levels of 72 patients, aged 6 days-33 months (median 113 days) and weight 2.4-13.2 kg (median 5.3 kg, Quartile (Q)1: 2.4-4.10, Q2: 4.2-5.2, Q3: 5.3-6.2, Q4: 6.2-13.2 kg ) were available for analysis comprising younger individuals compared to the patients in the Valkenburg et al. model (35 patients, aged 3-36 months (median age 177 days), weight 3.6-12.9 kg (median 6.14 kg). Generally, concentrations in the PACS study were lower than in the Valkenburg et al. study, as a result of overall lower morphine doses compared to the Valkenburg et al. study. The Valkenburg et al. model was able to predict the external data with minimal bias or trends, except for 1) a slight under-prediction in lowest age/weight quartile and a slight over-prediction in the highest age/weight-quartile in the population predicted concentrations for morphine concentrations, and 2) a general over-prediction of morphine-6-glucuronide concentrations, which could be related to the lower concentrations as a result of lower dosages. The NPDE analysis showed similar trends. ETA versus covariate plots revealed trends with both age and weight in the formation clearance of morphine-3-glucuronide and elimination clearance of morphine-3-glucuronide in children at the lowest (below 15 days of age) and highest age/weight-ranges.
Conclusions: The original model for morphine in children after cardiothoracic surgery as developed by Valkenburg et al., showed to predict well the morphine and metabolite data of patients in our study with similar age and weight of the Valkenburg et al. study. Slight deviations were observed when extrapolating the model to individuals with a lower and higher age and weight range.
References:
(1) Valkenburg AJ et al., Pediatric Critical Care Medicine. 2016 Oct 1;17(10):930–8.
(2) Krekels EHJ et al., Clinical Pharmacokinetics. 2011;50(1):51–3.
(3) Knibbe CAJ et al., Clinical Pharmacokinetics. 2009;48(6):371–85.