Orestis Papatryfonos1, Linda Wanika1, Amy Cheung2, Michael Chappell1
1School of Engineering, University of Warwick , 2Certara
Objectives: The main objective of this project was to develop a novel population Pharmacokinetic-Pharmacodynamic (PK-PD) model incorporating pain observations from the McGrath Colour Analog Scale (CAS), which assesses pain in children aged 6 to <12 years, receiving the opioid drug Tapentadol. The pain explored for this project was postoperative acute pain, measured on a 0 (no pain) to 10 (worst pain) scale [1]. Literature on PK-PD models utilising pain scale scores in paediatric populations is scarce, hence, this project hopes to provide a foundational step towards modelling the effects of Tapentadol in describing pain responses more accurately. This work forms part of the research undertaken within the EU project ERAMET for the development of an ecosystem for the rapid adoption of simulation methods that aims to create a credibility framework for orphan and paediatric drug development modelling and simulation. Methods: The analysis consisted of two stages: developing an appropriate PK model and coupling this with a corresponding PD model. The data utilised were obtained from the clinical trial NCT01729728 [1], focusing on postoperative acute pain. Tapentadol PKs were modelled using a linear one-compartment model, where the central compartment represents the plasma concentration of the drug. Attempts were made to incorporate a two-compartment model to account for the plasma concentration of metabolites, but this did not adequately describe the data and was thus not taken forward. The unknown parameters in the PK model were estimated using Monolix [2], and the resulting population estimates were fixed and then applied to the later stage of the analysis, where pain score was considered. The PK model was then coupled with a differential equation model describing the rate of change of pain intensity based on the CAS scores. The PD component followed an indirect Michaelis-Menten (MM) formulation, and its corresponding parameters were estimated in a population manner, again within Monolix. The maximum unbound systemic concentration (Imax) was fixed at 1, assuming a fully inhibiting model. This indirect model was chosen because opioid efficacy varies among children, unlike a direct MM model which assumes a more immediate drug effect. Results: The one-compartment PK model provided good fits to the observed drug concentration data, effectively capturing inter-individual variability. The estimated parameters included the absorption rate (Ka), clearance (CL), and volume of distribution (V). Several simulations were conducted, adjusting the initial parameter estimates for both the fixed effects and their standard deviations to achieve an RSE below 50%. The volume of distribution (V) was estimated at 430.86 L with an RSE of 5.94%, while clearance (CL) was 102.48 L/h with an RSE of 4.12%, both demonstrating strong precision. The absorption rate (Ka) was estimated at 2.48 h?¹, with a higher RSE of 28.6%, nearly an order of magnitude larger than the other parameters. These results are comparable with the parameter estimates reported in [3] for the PK model across multiple paediatric age groups (<18 years), supporting model validation. When the model is coupled with the CAS PD model, the population parameter estimates provided a good representation of pain response. The production rate of pain (Kin) was estimated at 490.83 h?¹ with an RSE of 10.2%, while the degradation rate of pain (Kout) was 133.01 h?¹ with an RSE of 10.6%, both sharing similar levels of precision. The half-maximal inhibitory concentration (IC50) was estimated at 140.83 ng/mL, with an RSE of 33.2%. A possible reason for this increased RSE value may be the high variability in observed CAS scores across individuals, likely due to frequent pain fluctuations at several time points. This contrasts with the observed plasma concentrations, where the PK model parameter estimates showed lower RSE values as each individual followed a similar trend. To ensure model robustness, further validation was performed, including an assessment of AIC and BIC model fit measures generated within Monolix, as well as an evaluation of the standard deviations of random effects. Conclusion: Overall, a novel PK-PD model has been developed to describe Tapentadol’s effect on postoperative acute pain in children aged 6 to <12 years, using the CAS. Model performance was evaluated based on parameter estimation accuracy, RSE assessment and statistical model fit criteria. Further work will focus on extending the PD model to better characterise the pain pathway in paediatrics.
[1] ClinicalTrials.gov, “NCT01729728: Pharmacokinetic, Efficacy and Safety Study of Tapentadol Oral Solution in Children With Postoperative Pain” [Online]. Available: https://clinicaltrials.gov/study/NCT01729728 [Accessed: 07-03-2025]. [2] LIXOFT, “Monolix, the best tool for Model Based Drug Development” [Online]. Available: https://lixoft.com/products/monolix/ [Accessed: 07-03-2025]. [3] E. Watson, A. Khandelwal, J. Freijer, J. van den Anker, C. Lefeber, and M. Eerdekens, “Population pharmacokinetic modeling to facilitate dose selection of tapentadol in the pediatric population,” J. Pain Res., vol. 12, pp. 2835–2850, 2019. doi: 10.2147/JPR.S208454
Reference: PAGE 33 (2025) Abstr 11614 [www.page-meeting.org/?abstract=11614]
Poster: Drug/Disease Modelling - Paediatrics