Thi Minh Nguyet Nguyen1, Budi Octasari Susanto1, Yanmin Hu2, Anthony Coates2, Ilse Dubbelboer1, Ulrika Simonsson1
1Department of Pharmaceutical Biosciences, Uppsala University, 2Institute for Infection and Immunity, St George's University of London
Introduction: In preclinical settings, the response of Mycobacterium tuberculosis to drug exposure can be observed through in vitro experiments such as time-kill assays (TKA). In a TKA, the bacterial response to treatment is studied longitudinally over multiple time points under natural growth conditions and at different drug concentrations. Performing TKA experiments can be lengthy, depending on the bacterial growth phase being measured. The stationary-phase is when the bacteria is in equilibrium and mimics the human infection more at the time of treatment. However, the experiments require more time than the log-phase culture where the bacteria is exponentially growing. The advantage of the latter is that it provides more rich information, such as information about the non-multiplying subpopulation. However, it remains unclear whether the stationary-phase TKA is necessary for developing antitubercular agents that do not affect non-multiplying bacteria. Objectives: The first aim of this work was to compare the identifiability of pharmacokinetic-pharmacodynamic (PKPD) relationships using only log-phase in vitro data with those derived from both log- and stationary-phase in vitro data. The second aim was to compare the model-based predictions of early bactericidal activity (EBA) using log-phase in vitro data and EBA predictions using both log- and stationary-phase in vitro data to the observed clinical EBA data. Methods: In vitro TKA data with M.tuberculosis H37Rv in log- and stationary-phases, in the absence of drug or in the presence of rifampicin or isoniazid, were used to estimate the PKPD relationships of rifampicin and isoniazid as monotherapy, using a multistate tuberculosis pharmacometrics (MTP) model [1]. The PKPD relationships derived from these in vitro-based MTP models for rifampicin and isoniazid in monotherapy were then integrated with translational factors, including population pharmacokinetics, mycobacterial susceptibility, bacterial growth phase, lung distribution model, and post-antibiotic effect parameters. A total of 1000 virtual subjects were simulated from the population pharmacokinetic model, incorporating inter-individual variability, inter-occasion variability, and the distributions of covariates (e.g., weight, height) to predict clinical EBA [2]. The EBA predictions were then compared to the observed clinical EBA data. The in vitro data were modeled in NONMEM 7.5.1, and EBA predictions were simulated using the “deSolve” (version 1.4) package in R (version 4.2.3). Results: Using only log-phase in vitro data, the PKPD relationships of rifampicin in the MTP model included effects on inhibiting the growth of fast-multiplying (F) subpopulation and killing both F and slow-multiplying (S) subpopulations. By using both log- and stationary-phase data, an additional killing effect on non-multiplying (N) subpopulation was captured. EBA predictions for rifampicin based on only the log-phase MTP model overestimated observed clinical data, whereas predictions aligned well with observed EBA data when both log- and stationary-phase data were included. For isoniazid, the results were consistent between using only log-phase in vitro data and both log- and stationary-phase data. The PKPD relationships of isoniazid included inhibition of F growth and killing of both F and S subpopulations. The EBA predictions for isoniazid were largely consistent with observed data across different studies, irrespective of the use of in vitro TKA data. Conclusion: Including both log- and stationary-phase TKA data is unnecessary for EBA predictions if the antitubercular drug does not affect non-multiplying bacteria. However, for drugs that do target non-multiplying bacteria, incorporating both log- and stationary-phase data is essential for accurate early clinical efficacy prediction, as exemplified in the case of rifampicin.
Poster Session III, Ιουν?ου 5, 2025, 9:50 πμ – 11:45 πμ [1] Clewe O, Aulin L, Hu Y, Coates ARM, Simonsson USH. A multistate tuberculosis pharmacometric model: A framework for studying anti-tubercular drug effects in vitro. J Antimicrob Chemother 2016; 71: 964–74. [2] Wicha SG, Clewe O, Svensson RJ, et al. Forecasting Clinical Dose-Response From Preclinical Studies in Tuberculosis Research: Translational Predictions With Rifampicin. Clin Pharmacol Ther 2018; 104: 1208 – 18.
Reference: PAGE 33 (2025) Abstr 11592 [www.page-meeting.org/?abstract=11592]
Poster: Drug/Disease Modelling - Infection