Karen Leys1, Dr. Agustos Ozbey2, Miao-Chan Huang1, Julia Macente1, Dr. Marina-Stefania Stroe3, MD PhD Timo de Haan4, MD Prof. Floris Groenendaal5, MD PhD Evelyne Jacqz-Aigrain6, MD Prof. Karel Allegaert7,8,9,10, Prof. Steven Van Cruchten3, MD Prof. Anne Smits8,10,11, Prof. Pieter Annaert1,12
1Drug Delivery and Disposition, Department of Pharmaceutical and Pharmacological Sciences, KU Leuven, 2Non-clinical development, Idorsia Pharmaceuticals, 3Comparative Perinatal Development, University of Antwerp, 4Department of Neonatology, Emma Children's Hospital, Amsterdam University Medical Center, 5Department of Neonatology, Wilhelmina Children's Hospital, University Medical Center Utrecht and Utrecht University, 6Hopital Robert-Debre AP-HP, Clinical Investigation Center, 7Department of Hospital Pharmacy, Erasmus MC, 8Department of Development and Regeneration, KU Leuven, 9Clinical Pharmacology and Pharmacotherapy, Department of Pharmaceutical and Pharmacological Sciences, KU Leuven, 10L-C&Y, KU Leuven Child & Youth Institute, 11Neonatal Intensive Care Unit, University Hospitals Leuven, 12BioNotus CommV
Introduction/Objectives: Therapeutic hypothermia (TH; core body temperature lowered to 33.5°C for 72 hours), is applied to neonates with moderate to severe hypoxic-ischemic encephalopathy after perinatal asphyxia, to decrease morbidity and mortality [1]. The impact of TH on drug metabolism is still poorly understood, and physiologically-based pharmacokinetic (PBPK) models in this population are not yet available [2]. The I-PREDICT project aims to fill this gap by developing a PBPK framework to describe and predict the impact of TH on drug disposition in this special population. The current study aims to report on the model development for the antiseizure drug midazolam (MDZ), administered intravenously (IV) during TH, and its primary metabolite, 1’-hydroxymidazolam (1’-OH-MDZ) in cooled neonates [2]. Methods: PK-Sim® version 11.3 was used as the modelling platform to develop the models simulating IV administration. For MDZ, the base model was established in the healthy adult population from the PK-Sim compound template, while 1’-OH-MDZ was adapted from the model reported by Johnson 2023 [3 ]. For the healthy adult population, 17 datasets were used for verification. After model verification in adults, the model was extrapolated to the paediatric population (6 months – 18 years old). Following the verification of the model for the paediatric population with 6 datasets, it was then further extrapolated to the neonatal populations. Two datasets were available for verification of the non-cooled neonatal model (1–28 days old) [4-5]. The PK-Sim default ontogenies were used, except for UGT1A4, for which the ontogeny profile of UGT1A1 was used, considering it is more biologically plausible [6]. The PBPK model was then further modified to reflect the ADME changes in the cooled population using in vitro data generated in-house based on suspended human adult hepatocytes (unpublished), which indicated that MDZ intrinsic clearance at 33.5°C was approximately 43% lower compared to 37°C. Furthermore, the fraction unbound was recalculated for a concentration of 22 mg/mL albumin (vs 29 mg/mL for non-cooled neonates) [7-8], and the glomerular filtration rate was decreased by 23.9% [2]. The cooled (1–5 days old) neonatal model was then verified using 2 datasets [9-10]. Results: The predicted plasma concentrations of MDZ were within 2-fold ratio of the AUC for all clinical dataset in adults, paediatrics (6 months – 18 years old), and the non-cooled neonatal population (1–28 days old). The developed MDZ models had an AAFE/AFE (absolute average fold error/average fold error) of 1.2/0.8, 1.1/1.1, and 1.2/1.1 or the healthy adult, paediatric, and non-cooled neonatal models, respectively. The non-cooled neonatal clinical studies were from NICU patients, which can explain some discrepancy in predictions. Nevertheless, incorporating the physiological changes occurring during TH, as identified in literature and in vitro experiments, improved the cooled neonatal model (1–5 days old), reducing the deviation from observations from 15-fold to 6-fold compared to using default healthy neonatal parameters. Conclusions: We developed a PBPK model and the results highlight the potential of PBPK to capture the changes in pharmacokinetics due to cooling caused by physiological changes and enzyme activity. The current understanding of the impact of TH on physiological parameters allows for improvement in predicting drug disposition in cooled neonates when the altered physiological parameters are available, and support dosing of midazolam in the future. However, the MDZ model needs further optimisation in the cooled neonatal population (1–5 days old). As a next step, the MDZ and 1’-OH-MDZ models will be further improved, first in non-cooled neonates with additional clinical data from the non-cooled but healthy neonatal population (1–28 days old). Then this will be extrapolated to the cooled neonatal population and further improved with optimization based on clinical data, e.g., adjusting for the cardiac output and incorporation of additional in vitro data (to be generated in human neonatal hepatocytes). Moreover, further improvements can be considered with additional in vitro experiments focussed on 1) proteomics to explore the temperature impact on the abundance of drug metabolising enzymes and transporters and 2) the impact of temperature on other metabolites.
[1] LaRosa et al. Front. Pediatr. 2017 [2] Leys et al. Expert Opin. Drug Met. 2023 [3] Johnson et al. Drug Metab. Dispos. 2023 [4] Jacqz-Aigrain et al. Eur. J. Clin. Pharmacol. 1992 [5] de Wildt et al. Clin. Pharm. Therap. 2001 [6] van Groen et al. Pharmacol. Rev. 2021 [7] Murinaman et al. Eur. J. Pediatr. 2017 [8] Han et al. Pharmaceutics 2025 [9] Favié et al. Neonatology 2019 [10] Welzing et al. Klin. Padiatr. 2013
Reference: PAGE 33 (2025) Abstr 11535 [www.page-meeting.org/?abstract=11535]
Poster: Drug/Disease Modelling - Paediatrics