Anh Quan Truong 1, Sai Dyviash 1, David Z. D’Argenio 2, Gauri G. Rao 1
1 USC Mann School of Pharmacy and Pharmaceutical Sciences (Los Angeles, United States), 2 Department of Biomedical Engineering, University of Southern California (Los Angeles, United States)
Objectives:
The continued spread of antimicrobial resistance demands an antibiotic development paradigm that explicitly accounts for the host immune response. This reality underscores the urgency of sustaining a continuous, robust antibiotic development pipeline. Standard neutropenic mouse models of infection such as thigh and lung infection models, common in new drug approval (NDA) submissions limit insight into the immune contribution to bacterial clearance and clinical outcomes [1, 2].
Mathematical modeling offers a powerful approach to characterizing the dynamics of the interplay between host immune responses and bacterial dynamics, enabling us to understand their interactions based on the host immune status. Existing mathematical models of bacterial dynamics often omit mechanistic interactions between cytokine signaling, granulopoiesis, and neutrophil–pathogen dynamics [3-13]. To address these gaps, we developed a proof-of-concept, physiology-based systems framework that integrates (1) human cytokine dynamics, (2) granulopoiesis, (3) bacterial growth kinetics, and (4) antibiotic pharmacokinetic/pharmacodynamic (PK/PD) during the course of infection. In this work, our first aim was to develop a human immunodynamic model linking cytokine signaling [11] to granulopoiesis [14], based on human endotoxemia data. Our second aim was to perform simulations to explore how host immune system characteristics (e.g., neutrophil production and maturation times) can influence bacterial clearance with and without antibiotic (meropenem) treatment.
Methods:
Host immunodynamics: We developed a mechanistic model of granulopoiesis and cytokine signaling during infection using data digitized from a human endotoxemia study. Four healthy male volunteers received an intravenous injection of Escherichia coli endotoxin (U.S. Reference strain O:113; lipopolysaccharide [LPS]) at a dose of 2 ng/kg body weight [15]. Data included interleukins (IL-1β, IL-6, IL-8, IL-10), tumor necrosis factor-α (TNF-α), granulocyte colony-stimulating factor (G-CSF), and neutrophil counts. Indirect response models with transit compartment models characterizing the delay with LPS induced cytokine production and feedback (e.g., IL-10 inhibition of pro inflammatory cytokines). Granulopoiesis model included a bone marrow (progenitors → metamyelocytes → band → segmented), blood (marginated and circulating pools), and target mediated G-CSF disposition compartments. The predicted neutrophil dynamics during LPS challenge were validated against three independent endotoxemia studies.
Antibacterial PK/PD: We developed a mechanism-based PK/PD model for meropenem against a clinical isolate of Pseudomonas aeruginosa (meropenem MIC = 4 mg/L) using static concentration time kill assay experimental data. Two bacterial subpopulations (susceptible and intermediate) and an Emax killing function characterized drug effect, with drug potency, EC₅₀ values estimated for each subpopulation.
Integrated systems-based framework: We coupled the host immunodynamic model and the mechanism-based antibacterial PK/PD model to form a system-based framework describing the dynamics of cytokine and neutrophils during pseudomonal pneumonia in the presence or absence of meropenem, at different immune statuses (immunocompetent and neutropenia). Monte Carlo simulations (n = 5000) were performed to model temporal interactions between neutrophils and bacteria and to identify key host characteristics that influence neutrophil-mediated bacterial clearance. We compared standard PK/PD index (%fT>MIC) based probability of target attainment (PTA) [16] with outcomes from the host–pathogen–drug model framework.
Results:
Cytokines and feedback: LPS-induced concentration-time profiles for pro- and anti-inflammatory cytokines were adequately described by indirect-response models with transit delays. Temporally, TNF-α peaked first in blood (mean transit time, MTT = 1.25 h), followed by IL-6 (MTT = 2.14 h), IL-8 (MTT = 2.16 h), and finally IL-10 (MTT = 2.32 h). IL-10 suppressed the production of TNF-α and IL-6, capturing counter regulatory anti inflammatory feedback.
Granulopoiesis and neutrophils: Under LPS challenge, G-CSF production increased with delay (MTT = 2.96 h). LPS increased neutrophil margination (MTT = 0.74 h). Chemokine IL-8 enhanced mobilization of band cells and segmented neutrophils from the bone marrow to blood and increased the rate of neutrophil extravasation. The integrated host immunodynamic model adequately captured emergency granulopoiesis during infection and accurately described neutrophil dynamics in the validation datasets.
Antibacterial dynamics: The mechanism-based PK/PD model with two P. aeruginosa subpopulations (susceptible and intermediate) best described time-kill assay data. EC₅₀ values: 17.0 mg/L (susceptible subpopulation) and 164 mg/L (intermediate resistant subpopulation).
Host-pathogen-drug simulations:
No treatment: Immunocompetent and mildly neutropenic individuals with pneumonia effectively cleared bacteria within ~24 h. However, moderate neutropenic produced bacterial regrowth with oscillations from delayed immune feedback. Severe neutropenic patients failed to clear infection, with bacterial grow achieving maximum levels based on the carrying capacity of the system.
With meropenem: A standard meropenem dosing regimen (1000 mg administered every 8 h) eradicated bacterial within 1 day irrespective of the host immune status with the integrated model framework. In contrast, conventional population PK based PTA calculation based-on %fT>MIC would not necessarily predict treatment success at MIC = 4 mg/L [16], highlighting the limitations of these PK/PD indices that only consider the drug activity while ignoring the host immune contributions.
Key host drivers: The basal neutrophil progenitor production rate was the dominant determinant of immune-mediated bacterial elimination; higher production rate improved bacterial clearance, regardless of immune status. The maturation time from marrow to blood for segmented cells has a bidirectional effect: in severe neutropenia, longer maturation correlated with higher clearance (reflecting complex feedback/mobilization dynamics), whereas in immunocompetent states, longer maturation reduced clearance.
Conclusions:
We present a proof-of-concept, translational systems framework that mechanistically links cytokine signaling, granulopoiesis, bacterial dynamics, and antibiotic PK/PD to quantify host–pathogen–drug interactions during pulmonary infection. By bridging in vitro bacterial killing data with human immunodynamics, the model elucidates how host physiological determinants govern bacterial clearance across immune states and exposes the limitations of traditional PK/PD indices. Simulations identified basal progenitor production rate and segmented-cell maturation time as critical levers of neutrophil-mediated clearance, offering testable hypotheses for dosing patients based on immune status. This integrative approach exemplifies Lewis Sheiner’s vision of connecting biology, biomarkers, and pharmacology to inform treatment regimen selection and design, and decision making in antimicrobial development.
References:
1. Craig, W.A., Clin Infect Dis, 1998. 26(1): p. 1-10; quiz 11-2.
2. Byrne, J.M., et al., Antimicrobial Agents and Chemotherapy, 2020. 65(1): p. 10.1128/aac.01983-20.
3. Li, Y., et al., Proceedings of the National Academy of Sciences, 2002. 99(12): p. 8289-8294.
4. Drusano, G.L., et al., Antimicrob Agents Chemother, 2010. 54(10): p. 4368-72.
5. Drusano, G.L., et al., Antimicrobial Agents and Chemotherapy, 2011. 55(6): p. 2693-2695.
6. Guo, B., et al., Antimicrob Agents Chemother, 2011. 55(10): p. 4601-5.
7. Malka, R., et al., J Clin Invest, 2012. 122(8): p. 3002-11.
8. Sadiq, M.W., et al., Journal of Pharmacokinetics and Pharmacodynamics, 2017. 44(2): p. 69-79.
9. Thorsted, A., et al., CPT Pharmacometrics Syst Pharmacol, 2023. 12(12): p. 1972-1987.
10. Saporta, R., 2025, Acta Universitatis Upsaliensis: Uppsala. p. 79.
11. Diep, J.K., T.A. Russo, and G.G. Rao, CPT Pharmacometrics Syst Pharmacol, 2018. 7(8): p. 507-516.
12. Ankomah, P. and B.R. Levin, Proceedings of the National Academy of Sciences, 2014. 111(23): p. 8331-8338.
13. Gjini, E. and P.H. Brito, PLoS Comput Biol, 2016. 12(4): p. e1004857.
14. Chen, W., et al., J Pharmacokinet Pharmacodyn, 2020. 47(2): p. 163-182.
15. de Kleijn, S., et al., PLoS One, 2012. 7(6): p. e38255.
16. Ehmann, L., et al., Int J Antimicrob Agents, 2019. 54(3): p. 309-317.
Reference: PAGE 34 (2026) Abstr 12199 [www.page-meeting.org/?abstract=12199]
Poster: Drug/Disease Modelling - Infection