Miriam S.R. Happ 1,2, Christin Nyhoegen 1, Nicole Zimmermann 1, Mara Baldry 3, Charlotte Costa 3, Delphine Cayet 3, Wilhelm Huisinga 2,4, Jean-Claude Sirard 3, Charlotte Kloft 1,2,5, Robin Michelet 1,5
1 Freie Universität Berlin, Institute of Pharmacy, Department of Clinical Pharmacy and Biochemistry (Berlin , Germany), 2 PharMetrX Graduate Research Training Program (Potsdam/Berlin, Germany), 3 Institut Pasteur de Lille, Center for Infection and Immunity of Lille, Inserm (Lille, France), 4 Institute of Mathematics, University of Potsdam (Potsdam, Germany), 5 shared senior authorship (, )
Introduction: According to recent WHO reports, 17% of bacterial infections worldwide in 2023 were caused by pathogens with antimicrobial resistance (AMR) [1], a global threat associated with 4.71 million deaths in 2021 [2]. With the expected rise in AMR related deaths [2], the need for solutions preserving availability of effective treatments and combating AMR development is increasing.
Combining immunomodulatory therapy with traditional antibiotics is a promising strategy against AMR. Immunomodulators have the potential to enhance efficacy of available antibiotics via synergism [3] and exhibit lower risk of resistance emergence by acting on the host instead of the pathogen [4-6]. The flagellin derivative FLAMOD locally stimulates innate immunity after nebulised administration with the aim to increase antibiotic efficacy in drug-resistant bacterial pneumonia [3,7,8]. Yet, drug-drug interactions with different antibiotics, the complex nature of immune responses and difficulties posed by low systemic bioavailability after inhalation, exclude the application of standard translational tools during the development of anti-infectives e.g., pharmacokinetic/pharmacodynamic (PKPD) indices, causing a translational gap for the development of FLAMOD. Additionally, individual preclinical experiment may provide only partial insights into the combination therapy e.g., the PK of one drug. Thus, to enable meaningful outcome translation and full understanding of the host-drug-drug-disease interplay (HDDDI), integrative approaches handling multi-source data are required.
Throughout the development of FLAMOD within the FAIR consortium, available data and mathematical models were integrated into a translational modelling and simulation (ModSim) platform [6], enabling translation of FLAMOD’s immunomodulatory effects across biomarkers and preclinical species, while applying the learn-predict-confirm paradigm. The simultaneous assessment of available models at each stage guided subsequent development steps, including dose selection for toxicology and first-in-human studies. While essential for biomarker selection and evaluation on FLAMOD mediated immune activity, thus far, this analysis was not able to relate immunoactivity to bacterial growth in the lung under combination therapy. Thus, we aimed to develop a comprehensive framework that characterises the full HDDDI in FLAMOD-antibiotic-therapy enabling relevant outcome prediction on bacterial growth.
Methods: A base model framework was developed by selecting representative elements outside of FLAMOD for each model component of the HDDDI i.e., amoxicillin as a representative antibiotic agent, a murine superinfection pneumonia model employing the amoxicillin-susceptible S. pneumoniae D39 (SPD39) strain (minimum inhibitory concentration (MIC)=0.016 µg/mL) [9] as base for the disease model, and CCL20 as biomarker for FLAMOD-mediated immune activation in the murine host.
Data from diverse preclinical mouse experiments were compiled into a unified dataset including PK data for both drugs, PD data on immune biomarkers and bacterial burden in the lung. Measurements of vehicle-, mono-, and combination treatment were included. PK and immune activation data were obtained from both healthy mice and the pneumonia model.
The base framework was developed in NONMEM 7.5.1, with each model component estimated separately (Laplacian algorithm) and parameter estimates fixed in subsequent steps. Beal’s M3 method [10] was applied to include data below the lower limit of quantification. Model performance was assessed based on OFV, AIC, parameter uncertainty, and visual predictive checks. After characterisation of each HDDDI component all models were integrated into one framework and jointly re-estimated. Subsequently, the base framework was expanded towards the amoxicillin-resistant S. pneumoniae 19F (SP19F) strain (MIC=1.5 µg/mL), by including MIC to account for antibiotic susceptibility. Lastly, the framework was used to explore different dosing scenarios by performing stochastic simulations (n=1000) accounting for residual variability.
Results: The base framework combined a published model of amoxicillin PK after intragastric (i.g.) administration and its effect on the logistic population growth of SPD39 in the lung [11] with a PKPD model describing the effect of FLAMOD on CCL20. The FLAMOD PKPD model described systemic and bronchoalveolar lavage (BAL) concentrations of FLAMOD and CCL20 after intranasal (i.n.) and intravenous administration. While no direct effect of FLAMOD or CCL20 on bacterial growth was observed, CCL20 concentrations were linked to the effect of amoxicillin, resulting in a maximum reduction of amoxicillin EC50 by 99.9%. A categorical impact of infection status on model parameters was evaluated at each development step, identifying increases of amoxicillin half-life, FLAMOD lung absorption and FLAMOD EC50 in superinfected animals. Joint parameter estimation caused an increase in parameter uncertainty compared to initial stepwise estimation but only minor deviation in parameter values.
Extension of the framework towards the resistant SP19F strain was achieved by considering MIC-normalised concentrations within the amoxicillin effect model. Inclusion of the additional data reduced parameter uncertainty (smaller RSEs) for most parameters, except for those related to bacterial growth. Stochastic HDDDI simulations showed that combining amoxicillin treatment (150 µg, i.g.) with FLAMOD (2.3 µg, i.n.) enhanced antibiotic effects in the susceptible SPD39, increasing the median log10-reduction at 12 h post-treatment (Δlog10_12h) by 1.77. In the resistant SP19F strain, higher antibiotic doses (1000 µg, i.g.) were required to achieve a comparable FLAMOD-mediated effect (Δlog10_12h = 1.54). Despite similar adjunct FLAMOD effects, total bacterial reduction in SP19F was only 5% and 30% of that in SPD39 for mono- and combination therapy, respectively.
Conclusions: A base framework, successfully integrating key HDDDI elements of FLAMOD–antibiotic combination therapy, was developed to enable the simulation-based assessment of treatment outcomes in a murine pneumonia model. The framework allows for longitudinal investigation of system components typically not assessed jointly in experiments e.g., FLAMOD PK and bacterial killing, and was extended to incorporate a resistant bacterial strain, demonstrating model robustness and broadening its application. Framework simulations gave critical insights into dose-outcome relations, such as the need for increased antibiotic doses to achieve comparable FLAMOD impact in resistant strains. Thus, it promotes a detailed understanding of the HDDDI, supporting decision-making in further clinical development of FLAMOD. The framework can be further improved by closing data gaps to reduce parameter uncertainties. Additionally, future work should expand the framework to other strains, antibiotics, biomarkers and species to further close translational gaps.
References:
Funding:
This project received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 847786.
1. World Health Organization Global antibiotic resistance surveillance report 2025: WHO Global Antimicrobial Resistance and Use Surveillance System (GLASS) – Summary. (2025).
2. Naghavi, M. et al. Global burden of bacterial antimicrobial resistance 1990–2021: a systematic analysis with forecasts to 2050. The Lancet 404, 1199–1226 (2024).
3. Matarazzo, L. et al. Therapeutic Synergy Between Antibiotics and Pulmonary Toll-Like Receptor 5 Stimulation in Antibiotic-Sensitive or -Resistant Pneumonia. Front Immunol 10, (2019).
4. Hancock, R. E. W., Nijnik, A. & Philpott, D. J. Modulating immunity as a therapy for bacterial infections. Nat Rev Microbiol 10, 243–254 (2012).
5. Scott, M. G. et al. An anti-infective peptide that selectively modulates the innate immune response. Nat Biotechnol 25, 465–472 (2007).
6. Michelet, R. et al. The use of translational modelling and simulation to develop immunomodulatory therapy as an adjunct to antibiotic treatment in the context of pneumonia. Pharmaceutics 13, 601 (2021).
7. FAIR HOME – FAIR Website. FAIR flagellin aerosol therapy at
8. Baldry, M. et al. Targeted delivery of flagellin by nebulization offers optimized respiratory immunity and defense against pneumococcal pneumonia. Antimicrobial Agents and Chemotherapy 68, e00866-24 (2024).
9. Mondemé, M. et al. Amoxicillin treatment of pneumococcal pneumonia impacts bone marrow neutrophil maturation and function. J Leukoc Biol 115, 463–475 (2023).
10. Beal, S. L. Ways to fit a PK model with some data below the quantification limit. Journal of Pharmacokinetics and Pharmacodynamics 28, 481–504 (2001).
11. Happ, Miriam S.R. et al. The impact of bacterial growth kinetics on predicted antibiotic efficacy under different pharmacodynamic model assumptions: An in vivo case example. In Thessaloniki – Greece PAGE 2025: Drug/Disease Modelling – Infection (Thessaloniki, Greece, 2025).at
Reference: PAGE 34 (2026) Abstr 12240 [www.page-meeting.org/?abstract=12240]
Poster: Oral: Preclinical and Translational modelling to support drug discovery and development