Ana RodrÃguez-Báez 1,2, Frederike Lentz 2, Benito GarcÃa 3, Ulrich Jaehde 1
1 Department of Clinical Pharmacy, Institute of Pharmacy, University of Bonn (Bonn, Germany), 2 Biostatistics and Special Pharmacokinetic Unit/Resear Division, Federal Institute for Drugs and Medical Devices (BfArM) (Bonn, Germany), 3 Department of Pharmacy, University Hospital (Léganes, Spain)
Introduction/Objectives: Clearance (CL) is a primary determinant of systemic exposure and plays a central role in dose selection and optimization. In first-in-child studies, doses can be defined using prior information from adults or children by applying allometric and maturation functions to extrapolate CL across age subgroups [1,2]. Body weight is occasionally normalized to the dataset’s median; however, in most paediatric pharmacokinetic analyses, scaling is standardized to a fixed adult weight (i.e., 70 kg). The impact of body weight normalization to 70 and 1 kg on the typical CL estimate (CLTV) has been studied and shown to have no effect on CLTV predictions [3,4]. However, normalization to a reference value not represented within the data, such as 70 kg in newborn studies, may lead to imprecise parameter estimates (e.g., CL) and thereby impact dose extrapolation. The aim of this study was to conduct an exploratory analysis to assess the impact of weight normalization on the precision of CL estimates in young paediatric patients.
Methods: Data were retrospectively collected from routine therapeutic drug monitoring records of gentamicin at University Hospital Severo Ochoa. Concentration-time data and associated demographic covariates were extracted for paediatric patients below 2 years. Model development and parameter estimation were performed using NONMEM version 7.5. Parameter estimation used the FOCE method with interaction. CL was scaled using a maturation function that characterizes glomerular filtration, with an age at which 50% of the mature value is achieved (TM50) of 47.7 weeks and a Hill coefficient of 3.4. Additionally, the primary pharmacokinetic parameters were scaled by fixed exponents of 0.75 and 1 for CL and volume of distribution, respectively. The impact of body weight (WT) normalization on CL estimation was evaluated using a reference scenario based on allometric scaling to the median WT per age group and was compared with three alternative scenarios: A) normalization to 70 kg, B) 1 kg, and C) median WT of the overall paediatric population. This evaluation was performed using a simulation-re estimation procedure (SSE), in which 300 trial replicates were simulated for each scenario. For evaluation, the relative standard error (RSE) and the relative root-mean-square error (RRMSE) were calculated.
Results: A total of 207 patients were included in the analysis. The median age of the patients was 127 days (range 20 to 702) with a median body weight of 6 kg (range 1.37 to 14.2). A two-compartment model with exponential inter-individual variability on CL and proportional residual unexplained variability best described the 355 gentamicin plasma concentrations. Convergence rates were generally high across scenarios (90%, 100%, and 97% for scenarios A, B, and C, respectively). Precision of CL estimates varied by normalization strategy and paediatric subgroup, with the highest RSE values observed in the term-newborn group. In this group, RSE ranged from 4.3 to 28.1% with 70 kg normalization, 4.1-20.6% with normalization to the overall paediatric median WT, and 4.1 to 15.6% with 1 kg normalization. The remaining subgroups, one-month, six-month, and one-year, exhibited lower RSE across normalization strategies with an overall range of <20%. These findings were consistent with the observed RRMSE values, which were within ≤5% of the true parameter value in the one-month group, and ≤10% in the six-month and one-year group.
Conclusions: Based on the SSE analysis, the accuracy of CL estimates decreased when the true median body weight was much lower and scaled to 70 kg, but slightly decreased as the weight distribution shifted towards higher values. CL scaled to an unrepresentative weight may lead to biased estimates that can be inappropriate for calculating optimal paediatric doses.
References:
[1] Manolis E et al. Pediatr Anesth 2011;21:214-221.
[2] van Rongen A et al. Expert Opin Drug Metab Toxicol 2022;18:99-113.
[3] Mahmood I et al. Clin Pharmacokinet 2017;56:1567-1576.
[4] Goulooze SC et al. Clin Pharmacokinet 2019;58:131-138.
Reference: PAGE 34 (2026) Abstr 12279 [www.page-meeting.org/?abstract=12279]
Poster: Drug/Disease Modelling - Paediatrics