Yu-Jou Lin1, Frances Okibedi1, Mats O. Karlsson1, Elin M. Svensson1,2
1Department of Pharmacy, Uppsala University, 2Department of Pharmacy, Radboud University Medical Center
Objectives: Bedaquiline (BDQ) has been shown to improve treatment outcomes in multidrug-resistant tuberculosis (TB) and is a key component in novel combination regimens. Exposure-response relationships for efficacy and safety and how these relate to long-term outcomes have been established under the regulatory-approved dosing regimen [1–4]. The once-daily BDQ-containing regimen is used in the approved bedaquiline-pretomanid-linezolid (BPaL) regimen, based on the results from the ZeNix trial [5]. However, this regimen lacks a higher loading dose in the first 2 weeks, which is important for maximizing the early bactericidal activity. An alternative BDQ daily dosing strategy, which keeps a 2-week high-loading dose, is currently being tested in the PARADIGM4TB trial (NCT06114628). Although some pharmacokinetic (PK) simulation studies have compared the PK profiles in different dosing regimens [6,7], whether the PK difference could be of clinical relevance regarding efficacy, safety, and long-term outcomes remains unclear. We aimed to build an interactive R Shiny simulation tool to fully evaluate BDQ dosing modifications in an integrated dose-PK-efficacy/safety-outcome modelling framework. Methods: Four models were incorporated in the framework: a population PK model characterizing BDQ and its metabolite M2 exposures [8], a PK-efficacy model [1], a PK-safety model [3], and a long-term TB treatment outcome model [4]. Within this framework, the PK-efficacy model described how the weekly average concentration of BDQ drove changes in mycobacterial load, and the PK-safety model characterized the effect of BDQ’s main metabolite M2 concentration on QT levels. In the outcome model, bacterial clearance and mycobacterial load derived from the PK-efficacy model at the end of treatment were identified as on-treatment predictors of conversion and recurrence. Other covariates in the framework included age, sex, race (black/non-black), baseline corrected calcium and potassium levels, baseline time-to-positivity in MGIT cultures, and time-varying weight and albumin concentration. The effect of drug resistance profiles on mycobacterial load decline was replaced by a half-life of mycobacterial load modifier. This approach enabled the models to account for variations in response across different resistance levels and background regimens at the time the models were developed, to allow for a more consistent evaluation of regimens in the current simulation. Three BDQ dosing regimens were evaluated: registered dosing (400 mg once daily [QD] for 2 weeks, then 200 mg thrice weekly [TIW] for 22 weeks), ZeNix dosing (200 mg QD for 8 weeks, then 100 mg QD for 16 weeks), PARADIGM4TB dosing (400 mg QD for 2 weeks, then 100 mg QD for 22 weeks). A large virtual population with certain covariate distributions representative of TB patients was generated using conditional distribution modeling [9], based on data from 556 patients pooled from three TB clinical trials available on the TB-PACTS platform [10]. Simulations were performed for 500 subjects sampled from the virtual population for each regimen using the mrgsolve package in R, and the results were visualized in the interactive Shiny application. Results: The median BDQ and M2 cumulative AUC were predicted 50% higher under the registered and PARADIGM4TB dosing at 2 weeks compared to the ZeNix dosing. The predicted proportion of patients with negative sputum culture results were comparable throughout the treatment period for the registered and PARADIGM4TB dosing, whereas the ZeNix dosing yielded up to 11% lower proportions during the first month. A similar trend was observed in the long-term outcome model, with the proportion of patients in the converted state reflecting the differences in negative sputum culture samples. However, proportions of patients experiencing recurrence or died remained similar across all regimens over 2 years. The predicted QTc intervals were between 376-442 ms (5th to 95th percentiles) for all the evaluated dosing strategies. Conclusions: The early treatment response differed between the PARADIGM4TB and ZeNix dosing strategies. The PARADIGM4TB dosing showed the advantages of earlier sputum culture conversion. Additionally, QT intervals were within safe limits across all regimens. This suggests that the PARADIGM4TB once-daily dosing is both efficacious and safe. The developed Shiny app can be used for future exploration of BDQ treatment under different scenarios, e.g., potential drug-drug interactions, and clinical trial simulations.
[1] Svensson and Karlsson, J Antimicrob Chemother (2017). [2] Tanneau et al., Br J Clin Pharmacol (2020). [3] Tanneau et al., CPT Pharmacomet Syst Pharmacol (2021). [4] Lin et al., J Antimicrob Chemother (2024). [5] World Health Organization. WHO operational handbook on tuberculosis: Module 5. (2022). [6] Salinger et al., Antimicrob Agents Chemother (2019). [7] Lin et al., Clin Pharmacol Ther (2024). [8] Svensson et al., CPT Pharmacomet Syst Pharmacol (2016). [9] Smania and Jonsson, CPT Pharmacomet Syst Pharmacol (2021). [10] Critical Path Institute. Access the TB-PACTS platform. https://c-path.org/ programs/tb-pacts/codr-database/.
Reference: PAGE 33 (2025) Abstr 11540 [www.page-meeting.org/?abstract=11540]
Poster: Drug/Disease Modelling - Infection