Wen Yao Mak 1, Jon Ulf Hansen 2, Natália Tassi 3, Bernhard Kerscher 3, Diego Vera-Yunca 1, Aghavni Ginosyan 1, Irene Loryan 1, Carina Vingsbo Lundberg 2, Lena E. Friberg 1
1 Department of Pharmacy, Uppsala University (, Sweden), 2 Bacteria, Parasites & Fungi, Statens Serum Institut (, Denmark), 3 Div. of Infectious Diseases, Paul-Ehrlich-Institut (, Germany)
Objectives:
The COMBINE consortium has developed a standardized murine pneumonia model to harmonize preclinical efficacy data and improve translational relevance in antimicrobial development¹. However, inter-laboratory variability may influence pharmacokinetic/pharmacodynamic (PK/PD) characterization and the derivation of antimicrobial exposure thresholds. This work evaluated the preclinical PK/PD characteristics of levofloxacin and meropenem and quantified laboratory-specific effects across independent laboratories.
Methods:
Levofloxacin (0.5-50mg/kg q8h s.c.) and meropenem (0.5- 300mg/kg q4h s.c.) were evaluated against three Klebsiella pneumoniae (DSM 116098, 116099, 30104) and three Pseudomonas aeruginosa (DSM 116110 (levofloxacin-resistant), 116114, 50071) strains, at two labs (Statens Serum Institut, SSI; Paul-Ehrlich-Institut, PEI), using the standardized COMBINE pneumonia model. All in‑vivo data were generated using an identical experimental protocol, and bioanalyses were performed at the same site using a validated LC–MS/MS analytical method. PK models were developed separately for each antibiotic using plasma and epithelial lining fluid (ELF) concentration-time data. Laboratory was tested as a covariate during model development, and changes in OFV guided covariate selection.
Three exposure metrics (fAUC/MIC, fCmax/MIC, %fT>MIC) were derived using laboratory-specific PK parameter estimates based on unbound plasma concentrations. Exposure-response relationships were evaluated per species and per laboratory using the neutropenic (induced by cyclophosphamide) murine data (n=478 across 6 strains). Sigmoidal-Emax models were fitted to 24-hour bacterial response (Δlog10 CFU) against each PK/PD index. Model performance was compared using AIC, residual unexplained variability (RUV), and R². Exposure targets corresponding to stasis, 1-log, and 2-log kill (defined as the PK/PD index value associated with model-predicted net bacterial change of 0, 1- and 2-log₁₀ change relative to the start of therapy) were derived, and compared between laboratories.
Results:
Levofloxacin PK was characterized by a four-compartment model with first-order absorption and linear elimination (ka=3.4 h-1, CL=8.8 L/h/kg), with pulmonary ELF distribution. RUV was 26% CV in plasma and 151% CV in ELF. Meropenem was characterized by a five-compartment model with first-order absorption and linear elimination (ka=2.8 h-1, CL=2.2 L/h/kg). Pulmonary distribution was described by two ELF compartments. RUV was higher for ELF (114% CV) than for plasma (81% CV).
For levofloxacin, SSI showed 67% higher relative bioavailability, with 42% and 50% lower ka and CL, respectively, resulting in 3.3-fold higher fAUCSSI compared with PEI. For meropenem, laboratory effects were significant on relative bioavailability (~3.5-fold higher at SSI), yielding a corresponding increase in fAUCSSI. Relative %fT>MIC ratios were dose-dependent (1.3- to 2.7-fold higher at SSI).
For levofloxacin against K. pneumoniae, fAUC/MIC demonstrated superior model performance, as indicated by AIC and RUV. Significant PK differences affected the subsequent exposure-response relationship, resulting in fold differences (ExposureSSI/ExposurePEI) of 0.7, 0.8, and 0.9 for stasis, 1-log and 2-log kill. For P. aeruginosa, exposure-response relationships differed more (fold difference: 0.2, 0.3, and 0.4). DSM 116110 did not achieve stasis at the highest tested dose (50 mg/kg), thus confirming resistance in vivo.
For meropenem, fAUC/MIC was the best index against K. pneumoniae based on the current data, with consistent exposure targets (fold lab differences: 0.7, 0.9, and 1.1 for stasis, 1-log, 2-log kill). Notably, for two SSI strains (DSM 116098 and 116099), Emax was not achieved at 300 mg/kg. The predicted PK/PD indices for P. aeruginosa were strain- and laboratory-dependent, with fAUC/MIC generally yielding improved statistical fits compared with %fT>MIC.
Conclusions:
The inclusion of laboratory as a covariate on both the levofloxacin and meropenem PK models suggested systemic and pulmonary PK differences across laboratories that impact exposure metrics and shift the exposure-response relationship. For levofloxacin, expressing antimicrobial effects as a function of exposure (i.e. fAUC) rather than dosage reduced inter-laboratory divergence and resulted in comparable PK/PD targets across laboratories. For meropenem, larger inter-laboratory differences were observed in EC₅₀ and ‘best’ target (fAUC/MIC or %fT>MIC). These findings highlight the importance of evaluating PK in each laboratory to ensure reliable target determination in preclinical infection models. Further evaluations will be performed using the General Model².
References:
¹Hansen JU, Arrazuria R, Hoover JL, et al. The COMBINE pneumonia model: a multicenter study to standardize a mouse pneumonia model with Pseudomonas aeruginosa and Klebsiella pneumoniae for antibiotic development. Microbiol Spectr. Published online January 14, 2026. doi:10.1128/spectrum.03464-25
²Karlsson MO, Molnar V, Bergh J, Freijs A, Larsson R. A general model for time-dissociated pharmacokinetic-pharmacodynamic relationship exemplified by paclitaxel myelosuppression. Clin Pharmacol Ther. 1998;63(1):11-25. doi:10.1016/S0009-9236(98)90117-5
Reference: PAGE 34 (2026) Abstr 11888 [www.page-meeting.org/?abstract=11888]
Poster: Drug/Disease Modelling - Infection