Pietro Laddomada 1,2, Alessandro Di Deo 1,2, Salvatore D'Agate 2, Marie Sylvianne Rabodoarivelo 3, Rebeca Bailo 3, Jordana Galizia 3, Dominique Ndjogou 4, Cyril Gaudin 4, Eik Hoffmann 4, Santiago Ramón-García 3,5,6, Oscar Della Pasqua 1,2
1 Consiglio Nazionale delle Ricerche (CNR) (Rome, Italy), 2 Clinical Pharmacology & Therapeutics Group, University College London (London, United Kingdom), 3 Department of Microbiology, Pediatrics, Radiology and Public Health, University of Zaragoza (Zaragoza, Spain), 4 Institute Pasteur Lille, Center for Infection and Immunity of Lille (CIIL) (, France), 5 Spanish Network for Research on Respiratory Diseases (CIBERES), Carlos III Health Institute (Madrid, Spain), 6 Research and Development Agency of Aragón (ARAID) Foundation (Zaragoza, Spain)
Introduction:The observed regrowth of Mycobacterium tuberculosis (Mtb) in vitro is often attributed to the emergence of phenotypical resistance or relapse-associated subpopulations, which enable the bacterium to survive killing effects even at very high concentrations[1][2][3]. However, distinguishing these is challenging due to a lack of validated activity markers [4], making the assessment of eventual subpopulations trivial[5]. This phenomenon may also be explained by overlooked factors, such as actual drug exposure.
In time-kill assays (TKA), nominal drug concentrations are assumed constant throughout the treatment period. However, drug instability and degradation are typically not accounted for, unless dedicated experiments are conducted to inform the actual antibacterial effect as a function of time and samples obtained for the measurement of actual drug concentrations [6].
Objectives: The aim of this work was to implement a model-based approach to assess the implications of drug degradation on bacterial regrowth and drug potency for bedaquiline (BDQ), pretomanid (PTD), and linezolid (LZD) under different experimental conditions and multiple markers of bacterial activity.
Methods: Data from TKA protocol, including exponential and stationary bacterial growth phases of Mtb strain H37rv in cholesterol media over a treatment period of 49 days, were available for this investigation. In vitro antibacterial activity, assessed as colony forming unit (CFU) and most probable number (MPN), was characterised over the following nominal concentration range: BDQ 0.061-10.51 µg/ml, PTD 1.25-20.0 µg/ml and LZD 16-32 µg/ml.
A two-compartment model was developed to characterise bacterial growth, PK/PD relationships, and drug effects. CFU data and the MPN–CFU difference were jointly modelled to represent solid and liquid media. The parameterisation captures differences in drug susceptibility across culture conditions, reflecting experimental influences on the metabolic phenotype of Mtb.
Logistic and Gompertz functions were evaluated to characterise bacterial growth dynamics. Bactericidal activity was described using a sigmoidal Emax model. In addition, different structural models were evaluated to describe the stability of BDQ, PTD, and LZD using data from drug stability experiments. The final degradation model for each compound was then integrated into the overall analysis to account for time-varying drug exposure. Finally, drug potencies were re-estimated taking into account this effect.
Results: The proposed two-compartment model was able to describe bacterial growth and drug effect on CFU, along with estimates of the difference between MPN and CFU. Bacterial growth dynamics were best described by a logistic model, and different growth rates were estimated for multiple experimental conditions (for stationary phase, 0.206 1/h and 0.223 1/h for CFU and MPN-CFU, respectively).
The concentration-effect relationship was described using a sigmoidal Emax model. Drug degradation over time was adequately described by a generalised linear model. Significant degradation was observed for BDQ and PTD (slopes -0.866 and -0.703, respectively), whereas LZD showed no statistically significant degradation.
Using nominal drug concentrationswithout drug degradation, EC50 and Emax estimates for BDQ were, respectively, 7.71 µg/mL and 1.76 1/h for CFU; 11.9 µg/mL and 1.16 1/h for MPN-CFU. Analogously, for PTD these estimates were 1.43 µg/mL and 1.19 1/h for CFU; 2.07 µg/mL and 1.27 1/h for MPN-CFU. The integration of the estimated linear function describing drug degradation led to an overall increase in drug potency, shifting the EC50 for the two markers to 1.35 µg/mL and 1.86 µg/mL for BDQ, to 0.53 µg/mL and 0.924 µg/mL for PTD. In both cases, incorporating the actual time-varying drug concentrations improved the overall goodness of fit of the model, enabling a more accurate characterisation of the bacterial load reduction and subsequent regrowth. Interestingly, LZD monotherapy demonstrated no antibacterial activity within the tested concentration range.
Conclusion: This study highlights the importance of an integrated model-based framework for the characterisation of bacterial growth dynamics and antitubercular activity against both replicating and non-replicating populations under different experimental conditions. Beyond describing the differences the antibacterial activity of the different drugs, our analysis demonstrates the importance of bioanalytical monitoring of concentrations to account for drug stability and degradation over the course of treatment in long term experimental protocols. By correcting for time-dependent changes in drug exposure, our findings show that bacterial regrowth may be driven by pharmacokinetic factors, rather than phenotypical resistance or tolerance. Such a comprehensive evaluation is essential for the selection of doses and dosing regimens as well as prioritisation of companion drugs for optimised combination therapies.
References:
[1] F. Boldrin, R. Provvedi, L. Cioetto Mazzabò, G. Segafreddo, and R. Manganelli, “Tolerance and Persistence to Drugs: A Main Challenge in the Fight Against Mycobacterium tuberculosis,” Aug. 26, 2020, Frontiers Media S.A. doi: 10.3389/fmicb.2020.01924.
[2] R. Aubry et al., “An improved PKPD modeling approach to characterize the pharmacodynamic interaction over time between ceftazidime/ avibactam and colistin from in vitro time-kill experiments against multidrug-resistant Klebsiella pneumoniae isolates,” 2023, doi: 10.1128/aac.00301-23.
[3] R. J. H. Hammond, F. Kloprogge, O. Della Pasqua, and S. H. Gillespie, “Implications of drug-induced phenotypical resistance: Is isoniazid radicalizing M. tuberculosis?,” Frontiers in antibiotics, vol. 1, 2022, doi: 10.3389/frabi.2022.928365.
[4] W. van Os and M. Zeitlinger, “Predicting antimicrobial activity at the target site: Pharmacokinetic/pharmacodynamic indices versus time–kill approaches,” Dec. 01, 2021, MDPI. doi: 10.3390/antibiotics10121485.
[5] M. Muliaditan and O. Della Pasqua, “Bacterial growth dynamics and pharmacokinetic–pharmacodynamic relationships of rifampicin and bedaquiline in BALB/c mice,” Br. J. Pharmacol., vol. 179, no. 6, pp. 1251–1263, Mar. 2022, doi: 10.1111/bph.15688.
[6] K. Mehta, T. Guo, P. H. van der Graaf, and J. G. C. van Hasselt, “Model-based dose optimization framework for bedaquiline, pretomanid and linezolid for the treatment of drug-resistant tuberculosis,” Br. J. Clin. Pharmacol., vol. 90, no. 2, pp. 463–474, Feb. 2024, doi: 10.1111/bcp.15925.
[7] This work has received support from the Innovative Medicines Initiatives 2 Joint Undertaking (grant No 853989).
Reference: PAGE 34 (2026) Abstr 12310 [www.page-meeting.org/?abstract=12310]
Poster: Drug/Disease Modelling - Other Topics