A general pharmacodynamic interaction modelling approach to assess semi-mechanistic synergy of ceftazidime/avibactam and fosfomycin in time kill experiments
Niklas Kroemer (1), Jean-Winoc Decousser (2), Patrice Nordmann (3), Sebastian G. Wicha (1)
(1) Dept. of Clinical Pharmacy, Institute of Pharmacy, University of Hamburg, Hamburg, Germany, (2) DYNAMYC Team – EA 7380, FACULTE DE SANTE, Université Paris-Est-Créteil Val-De-Marne, France, (3) Medical and Molecular Microbiology, University of Fribourg, Fribourg, Switzerland
Objectives: Detailed mechanistic understanding of in vitro pharmacodynamic (PD) drug interactions as provided by semi-mechanistic pharmacokinetic pharmacodynamic (PKPD) modelling is essential to be able to derive rational clinical antibiotic combination dosing regimens [1]. Subpopulation synergy or different types of mechanistic synergy can provide insights into PD drug interactions, but none is universally able to distinguish different interaction types and mechanisms [1][2][3]. Here, the properties of the general pharmacodynamic interaction model (GPDI) model could add additional benefits [1][4]. The aim of this study was therefore, to utilize the GPDI model as novel semi-mechanistic modelling approach in combination with subpopulation modelling to describe in vitro time kill data of ceftazidime/avibactam (CZA) and fosfomycin (FOS).
Methods: Time kill experiments of a clinical E. coli strain against CZA and FOS alone and in combination were conducted over 30 h and a pharmacometric metric model was developed in NONMEM 7.5.0. For description of the single drug effects sigmoidal maximum effect or slope models were tested. The combined drug effect was described by Bliss Independence and different base models with up to four different susceptible (S) and resistant (R) subpopulations with separate drug effects of CZA and FOF were tested [5]. Different implementations of the GPDI model on the (S) and (R) subpopulations were evaluated to describe the PD drug interactions. The GPDI model directionally identified perpetrator and victim drugs in PD interactions by the insertion of a GPDI-term. This GDPI term applies shifts on PD parameters like drug potency (EC50) or maximum drug effect (Emax) as result of a drug interaction. The model enables mono or bidirectional drug interactions with drugs being perpetrator and victim at the same time. The best model was chosen by the Akaike information criterion (AIC) [6].
Results: The final PD model included two bacterial subpopulations: an (S) population susceptible to both drugs and a corresponding (R) population with reduced susceptibility to both antibiotics. Drug effects of CZA and FOS on (S) and (R) were implemented as sigmoidal Emax models, except the effect of FOS on (S) was only supported by a slope model. Biological variability on the bacterial regrowth was implemented as interindividual variability with exponential variation on the inoculum of the resistant population. A model with CZA altering the potency of FOS on the (R) population as perpetrator outperformed a model with a vice versa interaction direction (AIC: 1439.9 and 1443.2). A bidirectional interaction was not supported by the data. Nevertheless, the model identified maximum shift of the potency of FOS of by 89% with a very high potency of CZA compared to the drug effect itself (EC50 of interaction: 0.001 µg/mL, drug EC50 on (R): 0.0852 µg/mL).
Conclusions: The novel approach implemented the GPDI model successfully as semi-mechanistic component in a subpopulation PD model for antibiotic time kill data. For the interacting drugs CZA-FOS, the model identified CZA as powerful perpetrator of the interaction enhancing the potency (EC50) of FOS on the (R) population. Therefore, the GPDI modelling approach added directional and quantitative information on the synergistic PD interaction of CZA and FOS. Prospectively, the model will be expanded as platform model to compare drug interactions in different strains. Thereby a scaling of the effect sizes with a strain depended susceptibility (i.e. the minimum inhibitory concentration (MIC)) as covariate is conceivable. Ultimately detailed mechanistic knowledge on the interactions could be utilized to derive highly efficacious clinical dosing regimens.
References:
[1] M. J. E. Brill, A. N. Kristoffersson, C. Zhao, E. I. Nielsen, and L. E. Friberg, “Semi-mechanistic pharmacokinetic–pharmacodynamic modelling of antibiotic drug combinations,” Clin. Microbiol. Infect., vol. 24, no. 7, pp. 697–706, 2018, doi: 10.1016/j.cmi.2017.11.023.
[2] C. B. Landersdorfer, N. S. Ly, H. Xu, B. T. Tsuji, and J. B. Bulitta, “Quantifying subpopulation synergy for antibiotic combinations via mechanism-based modeling and a sequential dosing design,” Antimicrob. Agents Chemother., vol. 57, no. 5, pp. 2343–2351, 2013, doi: 10.1128/AAC.00092-13.
[3] V. E. Rees, J. B. Bulitta, A. Oliver, R. L. Nation, and C. B. Landersdorfer, “Evaluation of tobramycin and ciprofloxacin as a synergistic combination against hypermutable pseudomonas aeruginosa strains via mechanism-based modelling,” Pharmaceutics, vol. 11, no. 9, 2019, doi: 10.3390/pharmaceutics11090470.
[4] S. G. Wicha, C. Chen, O. Clewe, and U. S. H. Simonsson, “A general pharmacodynamic interaction model identifies perpetrators and victims in drug interactions,” Nat. Commun., vol. 8, no. 1, p. 2129, Dec. 2017, doi: 10.1038/s41467-017-01929-y.
[5] C. I. Bliss, “The Toxicity of Poisons Applied Jointly,” Ann. Appl. Biol., vol. 26, no. 3, pp. 585–615, Aug. 1939, doi: 10.1111/j.1744-7348.1939.tb06990.x.
[6] H. Akaike, “A New Look at the Statistical Model Identification,” IEEE Trans. Automat. Contr., vol. 19, no. 6, pp. 716–723, Dec. 1974, doi: 10.1109/TAC.1974.1100705.
