Freek Relouw (1,2,3,4,5), Matthijs Kox (1), Rob Taal (2), Birgit Koch (3), Menno Prins (4,5,6), Natal van Riel (4,5)
(1) Department of Intensive Care Medicine, Radboud university medical center, Nijmegen, The Netherlands; (2) Department of Neonatal and Paediatric Intensive Care, Division of Neonatology, Erasmus University Medical Center, Rotterdam, The Netherlands; (3) Department of Hospital Pharmacy, Erasmus University Medical Center, Rotterdam, The Netherlands; (4) Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, The Netherlands; (5) Institute of Complex Molecular Systems, Eindhoven University of Technology, Eindhoven, The Netherlands; (6) Department of Applied Physics, Eindhoven University of Technology, Eindhoven, The Netherlands.
Introduction: Cytokines are inflammatory mediators that play a pivotal role in the initiation and orchestration of the immune response [1]. IL-6, a pro-inflammatory cytokine, is known to inhibit drug-metabolizing enzymes such as CYP3A4 [2]. In vitro and in vivo models of inflammation involve exposure to bacterial lipopolysaccharide (LPS), provoking an inflammatory response. These models provide insight into the dynamics of the response to inflammatory stimuli and sepsis pathophysiology, but there is still a gap between experimental and clinical settings [3]. In silico models have the potential to bridge this gap.
In silico models, initially based on animal data, have recently incorporated human subject data and accurately capture cytokine dynamics for bolus lipopolysaccharide (LPS) injections [4]. However, patients with infections are exposed to pathogenic stimuli for an extended period. The inflammatory response depends on the duration of the stimulus [5]. Unfortunately, current models are unable to correctly predict the response to longer LPS exposures [6].
Objectives:
- Develop an Ordinary Differential Equation (ODE) model to accurately model the inflammatory response to both bolus and continuous LPS infusion in humans.
- Bridge timescales between in vivo endotoxemia and clinical infections.
Methods: Interactions between model components were identified based on literature research. This multiscale model describes the changes in heart rate, temperature, blood pressure, and concentrations of TNF, IL-1β, IL-6, and IL-10 over time following exposure to LPS.
Model parameters were derived from dose-response curves obtained from in vitro experimental data found in the literature, wherever possible. The remaining parameters were estimated from in vivo calibration data. The calibration data consisted of the inflammatory response in healthy volunteers after bolus injections of either 1 (n=10) or 2 (n=15) ng/kg LPS [7]. Cytokines and vital signs were measured regularly for 8 hours after LPS administration.
Local- and multi-parameter sensitivity analyses (LPSA and MPSA) were conducted, followed by a (local) parameter identifiability analysis in the form of a profile likelihood analysis (PLA).
Validation was done by extrapolating the model, simulating the response to different LPS input profiles and comparing the model prediction with human experimental data from an independent validation dataset. This set consisted of a continuous 3-hour 1 ng/kg/h (n=10) infusion and bolus injections of 0.3 (n=10), 0.5 (n=6), 1 (n=6), and 2 (n=6) ng/kg of LPS [5] [7] [8].
Finally, the model was extrapolated to timescales beyond what is possible in a human in vivo model. These results were validated by comparing the steady-state biomarker levels in the simulation to those observed in sepsis patients in the ICU.
Results: The final ODE model consists of a system of 15 equations and 46 parameters. The LPSA and MPSA results were used to identify a subset of parameters to (re-)estimate in the calibration step. The PLA results confirmed that the model parameters could be uniquely estimated with the available calibration data.
The model described the calibration data well. Furthermore, the model prediction matched the non-linear inflammatory responses to previously unseen bolus injections of different LPS dosages and continuous infusion from the validation dataset. The model remained stable during the infection simulation and yielded cytokine levels in the ranges that are typically observed in patients in the ICU.
Conclusions: This in silico model of the inflammatory response represents a first step towards the development of an inflammation simulation model. It could be used and extended to better understand the disease mechanisms during inflammation and sepsis. The model presented here, akin to a systems biology framework, offers opportunities for expansion with more intricate pathways and interactions. Subsequently, it could be used for Quantitative Systems Pharmacology (QSP) approaches.
Moreover, the model allows for potential extensions to incorporate the dynamics of kidney and liver function [9]. By merging with a physiology-based pharmacokinetic model (PBPK), it holds promise for simulating diverse disease scenarios. Potentially facilitating the identification of potential biomarkers or covariates, shedding light on the interplay between inflammation and drug pharmacokinetics, and identifying opportunities for targeted therapies.
References:
[1] Zhang, J.-M. & An, J. Cytokines, Inflammation, and Pain. International Anesthesiology Clinics vol. 45 27–37 (2007).
[2] Harvey, R. D. & Morgan, E. T. Cancer, Inflammation, and Therapy: Effects on Cytochrome P450–Mediated Drug Metabolism and Implications for Novel Immunotherapeutic Agents. Clinical Pharmacology & Therapeutics vol. 96 449–457 (2014).
[3] van Lier, D., Geven, C., Leijte, G. P. & Pickkers, P. Experimental human endotoxemia as a model of systemic inflammation. Biochimie vol. 159 99–106 (2019).
[4] Dobreva, A. et al. A physiological model of the inflammatory-thermal-pain-cardiovascular interactions during an endotoxin challenge. The Journal of Physiology vol. 599 1459–1485 (2021).
[5] Taudorf S, Krabbe KS, Berg RM, Pedersen BK, Moller K. Human models of low-grade inflammation: bolus versus continuous infusion of endotoxin. Clin Vaccine Immunol 14, 250-255 (2007).
[6] Windoloski, K. A., Janum, S., Berg, R. M. G. & Olufsen, M. S. Characterization of differences in immune responses during bolus and continuous infusion endotoxin challenges using mathematical modelling. Experimental Physiology (2024).
[7] Kiers D, et al. Characterization of a model of systemic inflammation in humans in vivo elicited by continuous infusion of endotoxin. Sci Rep 7, 40149 (2017).
[8] Dillingh MR, et al. Characterization of inflammation and immune cell modulation induced by low-dose LPS administration to healthy volunteers. Journal of Inflammation 11, 28 (2014).
[9] Liu, F., Aulin, L. B. S., Manson, M. L., Krekels, E. H. J. & van Hasselt, J. G. C. Unraveling the Effects of Acute Inflammation on Pharmacokinetics: A Model-Based Analysis Focusing on Renal Glomerular Filtration Rate and Cytochrome P450 3A4-Mediated Metabolism. European Journal of Drug Metabolism and Pharmacokinetics vol. 48 623–631 (2023).
Reference: PAGE 32 (2024) Abstr 11263 [www.page-meeting.org/?abstract=11263]
Poster: Methodology - New Modelling Approaches