Estelle Chasseloup (1), Winn Ungphakorn (2), David McDougall (2), Bart Remmerie (1), Juan Jose Perez Ruixo (1), Anne-Gaëlle Dosne (1)
(1) Janssen Research & Development, (2) Parexel International, Australia
Introduction: Bedaquiline (BDQ) is a novel anti-mycobacterial agent approved in US, EU and other countries as part of combination therapy in patients (5 years old and above, >15 kg) with pulmonary multi-drug resistant Mycobacterium tuberculosis (MDR-TB). A population PK (popPK) model for BDQ was previously developed based on data from 9 clinical studies[1], and a parent-metabolite (BDQ-M2) model was developed on the same MDR-TB Phase 2 study data but without healthy volunteers (HVs) and single dose (SD) data[2].
Objectives:
- Perform an exploratory graphical analysis of BDQ and M2 in HVs and patients with tuberculosis (TB).
- Expand the previously developed model for BDQ by incorporating metabolite compartment(s) (CMT) to describe the observed BDQ and M2 PK data.
- Quantify popPK parameters, including typical parameter values, inter-individual variability, and residual variability.
- Identify and quantify covariate effects which contribute to the variability in the PK of BDQ and M2.
- Validation of the BDQ-M2 model on data from a Phase 3 study
Methods: The original dataset used to develop the popPK model for BDQ was expanded by including M2 data from the same studies: PK data from HVs recruited in 6 phase 1 studies and patients recruited in 3 phase 2 studies. The model building dataset consisted of 5222 BDQ concentrations and 4717 M2 concentrations from 479 adult participants (111 HVs and 368 patients with TB) with weight range of 30 kg – 113 kg. The data set for the external validation included 3087 BDQ concentrations and 3068 M2 concentrations from 282 adult MDR/Rifampicin-resistant-TB patients 15 years and older with weight range of 28-123.4 kg from the STREAM Stage 2 phase 3 trial.
An exploratory graphical analysis was conducted prior to the BDQ-M2 model development using the previously developed popPK model for BDQ as a starting point. The BDQ structure was reassessed, and the parameters were re-estimated after the inclusion of the M2 data. Covariates (i.e. race, population, study, and BDQ formulation) included in the prior model were retained, with additional covariates evaluated.
The external validation of the BDQ-M2 model was performed with the parameters fixed to the final estimates obtained for MDR-TB patients, using GOFs, VPCs, and individual profiles.
Results: The graphical analysis showed that BDQ exposure increased proportionally with dose and was higher in HVs than patients with MDR-TB. M2 exposure was comparable between the two populations but lower than BDQ (M2 peak following SD was about 100-fold lower). M2 had flatter profiles with almost no peak to trough fluctuations compared to BDQ. During the maintenance phase M2 had a higher accumulation (14-fold) than BDQ (2-fold).
The previously developed structural model for BDQ was retained with consistent re-estimated parameter values. A two-CMT model with linear input to the BDQ central CMT and linear elimination adequately described M2 data. To avoid identifiability issues, the central volume of distribution of M2 was fixed to BDQ and the FM parameter was estimated with a covariate effect to describe the differences in parent:metabolite ratios observed between HVs and patients, allowing BDQ to be eliminated as is and via metabolism to M2.
Typical apparent parameters for CLMET/Fm, clearance between the central and peripheral CMT for the metabolite (CLp1,MET/Fm), central and peripheral volumes of distribution for M2 (Vc,MET/Fm and Vp1,MET/Fm) were 4.95 L/L/h, 164L. The long half-lives of BDQ and M2 arise from their low clearance (2.62 and 4.95 L/h, respectively) and their large peripheral volume of distribution (7670 and 862 L for the largest CMTs, respectively).
The covariates included in the previous BDQ model (population and study effect on F, Black race and population on CLPAR/F and sex on Vc,PAR/F) remained statistically significant and affected also M2 PK. Black participants were predicted to have 31% lower BDQ and M2 AUC0-168 at Week 24 compared with non-black participants.
The validation showed that the popPK BDQ-M2 model provided an adequate description of the phase 3 data, despite a minor under-prediction of M2 at the population level.
Conclusion: A BDQ-M2 popPK model was developed using data from 9 phase 1 and 2 clinical studies and validated on phase 3 data. The model described the BDQ and M2 data in adult HV and patients with MDR-TB adequately, had good parameter precision, and can be used for simulation purposes.
References:
[1] McLeay SC, Vis P, van Heeswijk RPG, and Green B. The population pharmacokinetics of bedaquiline (TMC207), a novel anti-tuberculosis drug. Antimicrob Agents Chemother, Jun 2014.
[2] Svensson EM, Dosne AG, Karlsson MO. Population Pharmacokinetics of Bedaquiline and Metabolite M2 in Patients With Drug-Resistant Tuberculosis: The Effect of Time-Varying Weight and Albumin. CPT Pharmacometrics Syst Pharmacol. 2016 Dec;5(12):682-691. doi: 10.1002/psp4.12147. Epub 2016 Nov 8. PMID: 27863179; PMCID: PMC5192973.
Reference: PAGE 32 (2024) Abstr 11246 [www.page-meeting.org/?abstract=11246]
Poster: Drug/Disease modeling - Infection