Grace Shalom Govere 1,2, Jean-Michel Dogné 1,2, Happy Phanio Djokoto 1, Helene Haguet 1, Lisa Wellin 1, Adrien Olama 1, Lisa Hanquet 1, Camille Massaux 1, Reagan Kilela Songela 3, Flora T. Musuamba 1,2,3
1 University of Namur - Faculty of medicine (Namur, Belgium), 2 Belgian Federal Agency for Medicines and Health Products (Brussels, Belgium), 3 Université de Lubumbashi - Faculté de Sciences Pharmaceutiques (Lubumbashi, Democratic Republic of the Congo )
Objectives
Conducting pharmacokinetic (PK) studies in paediatric populations presents significant challenges due to physiological vulnerability, sampling limitations [1] and ethical considerations. [2] Sickle cell disease (SCD) is characterized by the presence of haemoglobin S (HbS) which polymerizes when deoxygenated, leading to the deformation of red blood cells and painful vaso-occlusive crises. [3] Therapeutic options for paediatric patients with SCD remain limited. [4]
HbS polymerization inhibitors bind to haemoglobin and increase its oxygen affinity, thereby reducing deoxygenation-induced HbS polymer formation and red blood cell sickling [5]. These agents represent a promising therapeutic class for the treatment of SCD. PK models for voxelotor, osivelotor and PF-07059013 were identified, and the feasibility of integrating them into a cross-product model to support paediatric PK study design was assessed.
The first objective of this study was to develop a cross-product population pharmacokinetic (PopPK) model to support the extrapolation of PK from adult to paediatric patients for HbS polymerization inhibitors. The second objective was to apply stochastic simulation-estimation (SSE) to optimize the paediatric PK study design for this therapeutic class.
Methods
Three PK models for voxelotor and the investigational agents osivelotor and PF-07059013 were implemented in NONMEM (Nonlinear Mixed Effects Modeling) software [6] from published literature. All models were two-compartment models developed for the characterization of PK in adult subjects.
Each model was subsequently adapted for paediatric extrapolation using allometric scaling exponents of 0.75 and 1.0 for clearance and volume of distribution respectively, and using maturation (ontogeny) functions. Exposure matching to safe and efficacious adult reference doses was performed based on AUC and Cmax. SSE using 500 replicated datasets was applied to design optimized paediatric PK studies for each drug based on the individual PK model. SSE was performed by comparing re-estimated parameter values to true parameter values.
Following the evaluation of structural features, parameterization, and predictive performance of the individual PK models, the feasibility of integrating the models into a cross-product model was assessed. The cross-product model was adapted for paediatric extrapolation using the same scaling exponents used for the individual models. SSE was applied based on the cross-product model to optimize paediatric PK study design for each drug. The resulting study designs were compared to the study designs derived from the individual PK models.
Results
The re-implementation of the voxelotor PK model was assessed by simulating concentration–time profiles for 279 subjects and comparing the resulting mean profile with the published mean concentration–time data to verify accurate model reproduction. The osivelotor model was adapted from the voxelotor model with osivelotor-specific concentration-time data for 6 subjects applied for model parameterization, whereas the PF-07059013 model was derived from a published preclinical semi-mechanistic model. Model adequacy was evaluated using goodness-of-fit diagnostics and visual predictive checks in R statistical analysis software to ensure accurate description of adult exposure prior to allometric scaling.
Shared structural features were identified across the three models, including a two-compartment plasma disposition framework. However, differences in PK linearity and in the relevance of target-mediated drug disposition (TMDD) modeling for PK were identified. These findings enabled the development of a cross-product model incorporating a shared two-compartment plasma base and drug-specific components.
Paediatric study designs were considered optimal when relative bias (RBias%) and relative root mean squared error (RRMSE%) for primary PK parameters were below 20% and 40%, respectively. The optimization of paediatric PK study designs using the cross-product model yielded study design parameters comparable to those obtained from individual models, including population size, cohort allocation strategy, number of samples per subject, and sampling frequency.
Conclusions
Although additional therapeutic options are needed for paediatric patients with SCD, ethical and practical constraints continue to limit extensive paediatric PK studies. Model-informed drug development represents a feasible strategy to generate reliable paediatric PK predictions while minimizing unnecessary paediatric exposure in clinical trials.
For HbS polymerization inhibitors, PK models can be integrated into a cross-product model consisting of a shared structural base and drug-specific components. Such a unified approach can support paediatric extrapolation and SSE-based study design optimization across multiple compounds within the therapeutic class, provided that drug-specific components are appropriately specified.
References:
[1] H. K. Batchelor and J. F. Marriott, “Paediatric pharmacokinetics: Key considerations,” Br. J. Clin. Pharmacol., vol. 79, no. 3, pp. 395–404, Mar. 2015, doi: 10.1111/bcp.12267.
[2] P. D. Joseph, J. C. Craig, and P. H. Y. Caldwell, “Clinical trials in children,” Br. J. Clin. Pharmacol., vol. 79, no. 3, pp. 357–369, Mar. 2015, doi: 10.1111/bcp.12305.
[3] M. Ferraresi, D. L. Panzieri, S. Leoni, M. D. Cappellini, A. Kattamis, and I. Motta, “Therapeutic perspective for children and young adults living with thalassemia and sickle cell disease,” Jun. 01, 2023, Springer Science and Business Media Deutschland GmbH. doi: 10.1007/s00431-023-04900-w.
[4] S. Chonat and C. T. Quinn, “Current standards of care and long term outcomes for thalassemia and sickle cell disease,” in Advances in Experimental Medicine and Biology, vol. 1013, Springer New York LLC, 2017, pp. 59–87. doi: 10.1007/978-1-4939-7299-9_3.
[5] R. M. Savic, M. L. Green, K. Jorga, M. Zager, and C. B. Washington, “Model-informed drug development of voxelotor in sickle cell disease: Population pharmacokinetics in whole blood and plasma,” CPT Pharmacometrics Syst. Pharmacol., vol. 11, no. 6, pp. 687–697, Jun. 2022, doi: 10.1002/psp4.12731.
[6] W. S. Park, “Pharmacometric models simulation using NONMEM, Berkeley Madonna and R,” Transl. Clin. Pharmacol., vol. 25, no. 3, pp. 125–133, Sep. 2017, doi: 10.12793/tcp.2017.25.3.125.
Reference: PAGE 34 (2026) Abstr 12191 [www.page-meeting.org/?abstract=12191]
Poster: Methodology - Study Design