A translational semi-mechanistic pharmacokinetic-pharmacodynamic framework to design animal studies: Application to linezolid and vancomycin
Diego Vera-Yunca, Lena E. Friberg
Department of Pharmacy, Uppsala University, Uppsala, Sweden
Objectives: One of the aims of the COMBINE project, which is part of IMI’s Antimicrobial Resistance (AMR) Accelerator programme, is to improve the translation of drug effects in terms of pharmacokinetics-pharmacodynamics (PK/PD) from preclinical to clinical development for antibacterial drugs. In this work, the aim was to use previously published linezolid (LZD) and vancomycin (VAN) PK and PK/PD models, developed on in vitro time-kill data, to facilitate the design of mouse studies, considering suggested therapeutic PK/PD targets as well as model-based strategies using predictions of bacterial concentrations.
Methods: Previously published LZD and VAN PK studies and models for mice and humans were identified [1–4]. The typical simulated exposures for a 70 kg person and for a 25 g mouse were coupled with a PK/PD model developed from in vitro S. aureus data by Wicha et al. [5] to predict bacterial concentrations at 24 h as colony-forming units (CFU)/mL. PK/PD indices were also correlated to the bacteria concentrations at 24 h. The initial bacterial inoculum was assumed to be 6.06 log10(CFU/mL), and minimum inhibitory concentrations (MIC) for LZD and VAN were set to the EUCAST breakpoints of 4 mg/L and 2 mg/L, respectively. The free AUC to MIC ratio (fAUC/MIC) was selected as the most relevant PK/PD index for both drugs based on existing literature. Model simulations, and data processing and plots were performed using mrgsolve and tidyverse R packages, respectively.
Results: Human simulations were performed using the recommended dosing regimens: 600 mg b.i.d. as an IV infusion of 1h (LZD) and 15 mg/kg b.i.d as an IV infusion with an infusion rate of 600 mg/h (VAN). Simulated control individuals (without drug) presented a bacterial concentration increase at 24h of 3.3 ∆log10(CFU/mL) at 24h. LZD resulted in bacteriostasis [∆log10(CFU/mL) = 0.3] with a fAUC/MIC of 27 where the clinical efficacy ranges between 55 and 83 [6], while VAN showed a more pronounced killing effect [∆log10(CFU/mL) = -5.6] with a fAUC/MIC of 55, where the clinical efficacy threshold is >180 [6]]. The dosing regimen in mice that matched the fAUC in humans for LZD at the recommended dose, 12 mg/kg IV bolus q12h, resulted in small bacterial growth (∆log10(CFU/mL = 1.1) while the 17 mg/kg q4h dose identified for VAN resulted in a modest killing effect compared to humans (∆log10(CFU/mL) = -0.9).
The dosing regimens for mice that showed the same bacterial killing effect as the one in humans with the recommended dosages were LZD: 18 mg/kg q12h, VAN: 40 mg/kg q4h. The corresponding PK/PD index target values were hence higher than reported for humans (LZD: fAUC/MIC = 42; VAN: fAUC/MIC = 133). In the case of VAN, the final total dose over 24h (240 mg/kg) was fractionated into different dosing intervals (2 h, 4 h, 6 h and 8 h) to test the relationship between fAUC/MIC and ∆log10(CFU/mL). The bacterial killing at 24 h changed depending on the selected dosing interval (from ∆log10(CFU/mL) = -5.6 if administered q2h to ∆log10(CFU/mL) = 2.7 if administered q8h). This shows that VAN’s bactericidal effect might not be only concentration-dependent, but also time-dependent. For LZD, ∆log10(CFU/mL) remained comparable across dosing regimens.
Conclusions: A PK/PD modelling framework was applied for LZD and VAN by coupling models from different stages of the drug development process: in vitro PK/PD models, in vivo PK models and clinical PK models. Exposure matching to translate human dosage to mouse dosage resulted in different degree of bacterial killing, despite that both drugs have been suggested to have an AUC-dependent effect. These results show that when designing preclinical experiments, using approved drugs as comparator or in exploring new drug combinations, PK/PD models that include bacterial dynamics and antimicrobial resistance have important explanatory value. As earlier indicated for other antibiotics, the PK/PD index approach may not hold to translate between mice and humans. In this study it was apparent that the fAUC/MIC target for bactericidal effects was dependent on the PK of VAN. The study supports the need for bacterial PKPD models to describe and predict the effect of current and new antibiotics across different species.
This work has received support from the EU/EFPIA Innovative Medicines Initiative 2 Joint Undertaking (COMBINE grant n° 853967).
- Aljutayli A, Thirion DJG, Bonnefois G, Nekka F. Pharmacokinetic equations versus Bayesian guided vancomycin monitoring: Pharmacokinetic model and model-informed precision dosing trial simulations. Clin Transl Sci. 2022 Feb 15;00(1):1–12.
- Bigelow KM, Deitchman AN, Li SY, Barnes-Boyle K, Tyagi S, Soni H, et al. Pharmacodynamic Correlates of Linezolid Activity and Toxicity in Murine Models of Tuberculosis. J Infect Dis. 2021 Jun 1;223(11):1855.
- Minichmayr IK, Schaeftlein A, Kuti JL, Zeitlinger M, Kloft C. Clinical Determinants of Target Non-Attainment of Linezolid in Plasma and Interstitial Space Fluid: A Pooled Population Pharmacokinetic Analysis with Focus on Critically Ill Patients. Clin Pharmacokinet. 2017 Jun 17;56(6):617–33.
- Kim HK, Choi SM, Kang G, Park KH, Lee DG, Park WB, et al. Comparison of In Vivo Pharmacokinetics and Pharmacodynamics of Vancomycin Products Available in Korea. Yonsei Med J. 2020 Apr 1;61(4):301–9.
- Wicha SG, Huisinga W, Kloft C. Translational Pharmacometric Evaluation of Typical Antibiotic Broad‐Spectrum Combination Therapies Against Staphylococcus Aureus Exploiting In Vitro Information. CPT Pharmacomet Syst Pharmacol. 2017 Aug 1;6(8):512.
- Abdul-Aziz MH, Alffenaar JWC, Bassetti M, Bracht H, Dimopoulos G, Marriott D, et al. Antimicrobial therapeutic drug monitoring in critically ill adult patients: a Position Paper. Intensive Care Med. 2020 Jun 7;46(6):1127–53.