Translation of the in vitro antimicrobial activity of afabicin against Staphylococcus aureus to in vivo bacterial killing dynamics using a pharmacokinetic-pharmacodynamic model based on time-kill experiments
Raphaël Saporta (1), Elisabet I. Nielsen (1), Valérie Nicolas-Métral (2), David R. Cameron (2), Annick Menetrey (2), Lena E. Friberg (1)
(1) Department of Pharmacy, Uppsala University, Sweden (2) Debiopharm International SA, Lausanne, Switzerland
Introduction: Afabicin is a prodrug of afabicin desphosphono, a first-in-class antibiotic. Afabicin desphosphono is an enoyl-acyl carrier protein reductase (FabI) inhibitor, a mode of action that targets fatty acid synthesis in Staphylococcus spp. [1]. Afabicin is in phase II development for the treatment of bone and joint infections due to staphylococci, and staphylococcal acute bacterial skin and skin structure infections [2]. Modelling bacterial dynamics during afabicin desphosphono exposure in vitro may contribute to better understanding of its antimicrobial effects and could be used to explore the translation from in vitro to in vivo bacterial killing.
Objectives: To characterize the pharmacokinetics-pharmacodynamics (PKPD) of afabicin desphosphono in a model based on data from in vitro time-kill experiments for various S. aureus strains, and to explore the translational capacity of the model for treatment efficacy in an in vivo mouse infection setting.
Methods: Data were gathered from 8 static in vitro time-kill studies. In these studies, a total of 21 S. aureus strains, with afabicin desphosphono minimum inhibitory concentrations (MIC) ranging from 0.004 to 0.03 µg/ml, were exposed to afabicin desphosphono concentrations ranging from 0.004 to 1 µg/ml. Colony forming unit (CFU) counts were determined over 48 hours.
A PKPD model was developed using NONMEM 7.5.0 and PsN 5.2.6. The base model assumed that bacteria could be in a drug-susceptible and growing state (S) or a resting state (R) [3]. To describe the drug effect, slope models or (sigmoidal) maximum effect (Emax) models were evaluated. Additional model components were evaluated: compartments to describe observed regrowth [4], a delay in drug effect through an effect compartment or lag function, a mixture model for subpopulations with different fraction of bacteria starting in the R state, and a scaling of the drug effect by MIC to account for strain differences.
A previously developed mouse pharmacokinetic (PK) model for afabicin [5] was used to simulate the PK after administration of different afabicin intraperitoneal doses. The unbound plasma concentration was used to drive the drug effect in the developed PKPD model to simulate in vivo bacterial killing in neutropenic mice. Simulations were performed with the mrgsolve package [6] in R.
Results: In total, 162 individual time-kill curves (867 CFU counts) were available for model building. The final model used a sigmoidal Emax model for drug effect, where differences in effect between strains were incorporated by scaling EC50 according to the strains’ MIC values (Emax=4.1 h-1, Hill coefficient=0.36, EC50=MIC0.16 µg/ml). Additional compartments inducing a reversible concentration-driven reduction of Emax (maximum possible reduction of 54%) were able to describe the observed regrowth. The other evaluated model components were not supported by the data. Visual predictive checks suggested a good description of the bacterial counts over time across all afabicin desphosphono concentrations studied in vitro.
Simulations of in vivo bacterial dynamics following administration of different afabicin regimens led to adequate predictions of bacterial count 24 hours after start of treatment when compared to observed bacterial counts (+/- 1 log10 difference).
Conclusions: The developed in vitro PKPD model successfully described the effect of afabicin desphosphono over time against 21 S. aureus strains in a 250-fold concentration range. The model was able to estimate differences in drug effect between strains with different MICs. Furthermore, predictions of in vivo bacterial counts suggested that the PKPD model can be used for translation purposes. The model’s ability to describe in vivo bacterial dynamics will be refined with additional modelling on data from mouse infection models. The model will also be further developed to explore the impact of immune response on in vivo PKPD.
References:
[1] Kaplan N, Albert M, Awrey D, et al. Mode of action, in vitro activity, and in vivo efficacy of AFN-1252, a selective antistaphylococcal FabI inhibitor. Antimicrob Agents Chemother. 2012;56(11):5865-5874.
[2] Wittke F, Vincent C, Chen J, et al. Afabicin, a First-in-Class Antistaphylococcal Antibiotic, in the Treatment of Acute Bacterial Skin and Skin Structure Infections: Clinical Noninferiority to Vancomycin/Linezolid. Antimicrob Agents Chemother. 2020;64(10):e00250-20.
[3] Nielsen EI, Viberg A, Löwdin E, Cars O, Karlsson MO, Sandström M. Semimechanistic pharmacokinetic/pharmacodynamic model for assessment of activity of antibacterial agents from time-kill curve experiments. Antimicrob Agents Chemother. 2007;51(1):128-136.
[4] Mohamed AF, Nielsen EI, Cars O, Friberg LE. Pharmacokinetic-pharmacodynamic model for gentamicin and its adaptive resistance with predictions of dosing schedules in newborn infants. Antimicrob Agents Chemother. 2012;56(1):179-188.
[5] Bader JC, Lakota EA, Bravo J, Dieppois G, Nicolas-Métral V, Miesel L, Lin KY, Ambrose PG, Bhavnani SM. Pharmacokinetic-Pharmacodynamic Analyses for Debio 1450 (Afabicin), a Staphylococcal-Specific Antibiotic, Using Data from a Murine-Thigh Infection Model. American Society for Microbiology Microbe 2017, New Orleans, LA.
[6] Baron K (2022). mrgsolve: Simulate from ODE-Based Models. R package, https://github.com/metrumresearchgroup/mrgsolve.