Niklas Hartung1, Gaby Wangorsch2, Wilhelm Huisinga1, Karin Weisser2
1University of Potsdam, 2Paul-Ehrlich-Institut
OBJECTIVES: Due to its toxic effects on bone, brain and liver at elevated exposure levels, a detailed understanding of Aluminium (Al) toxicokinetics is crucial for risk assessment of Al exposure from various sources like food, water, or medicinal products [1]. Recently, we established a mixed-effects physiologically-based toxicokinetic (PBTK) model for Al, built on a comprehensive set of toxicokinetic 26Al radiotracer data (totalling 938 data points in rats and human adults, measured in plasma, urine and six tissues), which could describe Al biokinetics in rats and human adults after single oral (p.o.) and intravenous (i.v.) doses of soluble Al salts [2]. However, several important aspects for the simulation of Al exposure were still lacking in this model, in particular (1) the extension to children, (2) a bone submodel accounting for age-dependent bone remodelling processes, and (3) incorporation of 27Al background exposure. The aim of this work was to extend our previous model in all these aspects to cover realistic scenarios, and to validate the resulting predictions in a diverse spectrum of Al administration schemes based on literature 26Al / 27Al exposure data [3]. METHODS: A consistent age-dependent physiological parametrization was obtained by interpolating reference values [4], combined with a specific model for GFR that includes maturation effects in neonates [5] and a new physiological bone submodel. In this submodel, Al first-order uptake into bone is assumed to be proportional to calcium (Ca) uptake, and release from bone identical to that of Ca. The Ca kinetic functions captured the age-dependent changes in bone turnover during childhood (puberty peak) and were obtained by consolidating a large number of published studies [6]. To deal with 27Al background exposure, uptake via food and initial levels at birth were incorporated into the simulations. The literature was screened for 26Al / 27Al plasma and tissue data after exposure from food (human reference values), parenteral nutrition (toxic levels in children and adults), and whole-body retention. The extended Al PBTK model had 25 estimated parameters (10 for fixed effects, 7 for random effects, 8 for the error model). It was encoded in the Mlxtran language and estimated using Monolix 2019R2 [7], based on the 26Al data considered previously [2]. Subsequently, simulations for the newly compiled 26Al / 27Al validation data were performed using R package mlxR version 4.2.0 [8]. These model predictions were then compared to the published data. The accuracy of this external evaluation was assessed visually and by calculating the fraction of data points in model-predicted ranges (90% confidence level). RESULTS: We have substantially extended our prior Al PBTK model to render it suitable for simulation of Al exposure from food, oral and parenteral medicinal products in humans from 0-50 years of age. A total of 19 studies suitable for model validation were identified, mainly reporting 27Al measurements, and all Al concentration data were digitized [9]. Al levels were predicted remarkably well after exposure from various sources and administration schemes (i.v., p.o.), on different timescales (days to years) and for different age groups (newborns, children, adults). Enteral Al exposure from food was accurately predicted for plasma and all toxicologically relevant tissues (bone, liver and brain), both in terms of central trends and inter-individual variability. Al exposure under parenteral nutrition was within the predicted range for 13 of 14 tissue data points in toxicologically renevant tissues and 24 of 37 plasma / urine samples, over different administration schedules and in both children and adults. The new bone submodel also substantially improved the agreement with previously considered 26Al long-term whole-body retention data, which were overpredicted in our previous Al PBTK model from 1 week onwards, but accurately described up to 10 years after dosing using the extended Al PBTK model. CONCLUSIONS: We have developed the first Al PBTK model in humans which includes a bone-seeking metal-specific bone submodel. We consider it valid for simulations of Al exposure in humans from birth to adulthood from various sources.
[1] Krewski D et al. J. Toxicol. Environ. Health B 10(sup1):1-269, 2007 [2] Hethey C et al. Arch. Tox. 95(9):2977-3000, 2021 [3] Hartung N et al. Arch. Tox., 2025 (accepted) [4] Valentin J et al. Ann. ICRP 32(3-4):1–277, 2002 [5] Rhodin M et al. Pediatr. Nephrol. 24(1):67-76, 2009 [6] Hartung N et al. Bone 189:117254, 2024 [7] Monolix version 2019R2. Antony, France. Lixoft SAS, 2019 [8] Lavielle M et al. (2021). mlxR: Simulation of Longitudinal Data (R package version 4.2.0). [online] https:// CRAN.R-project.org/package=mlxR [9] Rohatgi A (2022) WebPlotDigitizer [online] https://automeris.io/WebPlotDigitizer
Reference: PAGE 33 (2025) Abstr 11466 [www.page-meeting.org/?abstract=11466]
Poster: Drug/Disease Modelling - Absorption & PBPK