Linda B.S. Aulin1, Robin Michelet1, Moreno Ursino2,3,4, Sandrine Boulet3,4, Jean-Claude Sirard5, Emmanuelle Comets6,7, Sarah Zohar3,4, Charlotte Kloft1
1 Freie Universitaet Berlin, Institute of Pharmacy, Dept. of Clinical Pharmacy & Biochemistry, Berlin, Germany 2 Unit of Clinical Epidemiology, Assistance Publique-Hôpitaux de Paris, CHU Robert Debré, INSERM CIC-EC 1426, F-75019 Paris, France 3 INSERM, Centre de Recherche des Cordeliers, Sorbonne Université, Université Paris Cité, F-75006 Paris, France 4 Inria, HeKA, F-75015 Paris, France 5 Institut Pasteur de Lille, INSERM U1019, Centre d'Infection et d'Immunité de Lille, France 6 INSERM, Univ Rennes, EHESP, Irset (Institut de recherche en santé, environnement et travail) - UMRS 1085, F-35000 Rennes, France 7 INSERM, Université Paris Cité, IAME, F-75018 Paris, France
Background and Objectives:
In this work, we outline how mathematical modelling can be utilised throughout the development of an inhaled immunomodulatory drug candidate, FLAMOD. FLAMOD is intended for adjunct treatment of bacterial pneumonia and enhances the epithelial innate immune defence in the airways by stimulation of Toll Like Receptor 5 (TLR5), potentially improving therapeutic outcome. With this treatment strategy, the FAIR consortium [1] aims to address the increasing treatment failures related to antimicrobial resistance. Here, we describe a workflow leveraging in vitro and in vivo preclinical data to support the design of preclinical pharmacokinetic (PK), efficacy, and toxicity studies, and a first-in-human (FIH) trial.
Methods:
Our approach combines physiologically-based PK (PBPK), population PK, and pharmacodynamic (PD) modelling to amalgamate data and translate from different model systems, aiming to inform dose selection and study design. We developed a population PK model based on previously published mice and non-human primate (NHP) data, where the animals were given an intramuscular (im) dose of a compound similar to FLAMOD. The developed model was used to develop a rational PK sampling scheme of FLAMOD after intravenous (iv) administration in NHPs. The collected FLAMOD PK data were then used to build a PBPK model, which was developed using a middle-out-approach.[JCS1] The PBPK model was applied to obtain predictions regarding target lung doses for a subsequent PK/PD study of inhaled FLAMOD in NHPs.
FLAMOD efficacy was assessed using TLR5-stimulation markers. The efficacy data were obtained in human airway in vitro models as well as in mice and NHP studies.[JCS2] The dose-responses were modelled using sigmoidal Emax models. From each of these models, the Minimum Anticipated Biological Effect Level (MABEL) dose was calculated based on the dose resulting in 20% of the predicted maximum pharmacological effect (ED20). For the in vitro experiments, the human equivalent lung dose (HELD) was extrapolated using two different approaches; i) the surface area of the wells scaled to the surface area of the human lung [2], and ii) the amount of epithelial cells per well scaled towards the number of epithelial lung cells in the human lung [2]. For the in vivo experiments, allometric scaling of the lung surface area was used. The final human equivalent MABEL dose was selected based on maximizing the number of marker and studies the selected HELD would fall within the 95% confidence interval of the ED20. The same approach was used to obtain the HELD equal to the 90% of the predicted maximum pharmacological effect (ED90). The ED90 was used as a surrogate for the No Observed Adverse Effect Level (NOAEL), and applied according to FDA guidelines to determine inhaled the FLAMOD doses to be applied in toxicology studies in rats and beagle dogs. Furthermore, a proposal maximum recommended starting dose (MRSD) for a FIH trial was determined based on an anticipated safety window based on the MABEL-based and NOAEL-based dose.
Results and discussion:
A one-compartment model with a muscle-depot compartment captured im administration of a FLAMOD-like compound well in both mice and NHP. A model informed PK sampling schedule was derived and used to define a relevant sampling window between 0.01-3.5 h post iv dose of FLAMOD, updating the original experimental design significantly.
Sigmoidal dose-response behaviour was observed for the majority of the markers. Based on the combination of the in vitro and in vivo data, the extrapolated HELD for which most markers fell within the 95% CI of ED20 and ED90 were successfully determined and used to give design input for both the preclinical toxicology study and FIH trial.
In a second generation of our modelling framework, the developed dose-response and PK models will be linked to obtain dose-exposure-response models for FLAMOD, which can be updated based on the FIH trial and be used to design future proof-of-concept and dose-finding studies.
Conclusions:
In this project, we have shown how to apply different mathematic modelling techniques throughout preclinical drug development to inform study design and dose selection. As a next step, our modelling framework will be combined with a Bayesian framework to determine the dose escalation scheme. Furthermore, all models will continuously be applied and updated as preclinical and clinical data becomes available throughout the development of FLAMOD.
References:
[1] The ‘Flagellin Aerosol therapy as an Immunomodulatory adjunct to the antibiotic treatment of drug-Resistant bacterial pneumonia’ (FAIR) project; https://cordis.europa.eu/project/id/847786
[2] Robert R. Mercer et al. Cell Number and Distribution in Human and Rat American journal of respiratory cell and molecular biology 10.6 (1994): 613-624.)
Reference: PAGE 30 (2022) Abstr 10035 [www.page-meeting.org/?abstract=10035]
Poster: Drug/Disease Modelling - Infection