Silvia Grandoni (1,2), Peter Velickovic (1,2), Paul Healy (1,2), Salvatore D’Agate (1,2), Candie Joly (3), Julien Lemaitre (3), Laura Via (4,5), Oscar Della Pasqua (1,2)
(1) Consiglio Nazionale Delle Ricerche (CNR), Rome, Italy, (2) Clinical Pharmacology and Therapeutics Group, University College London (UCL), London, United Kingdom, (3) Université Paris-Saclay, Inserm, CEA, Center for Immunology of Viral, Auto-Immune, Hematological and Bacterial Diseases (IMVA-HB/IDMIT), Fontenay-aux-Roses, France. (4) Tuberculosis Research Section, Laboratory of Clinical Infectious Diseases, and (5) Tuberculosis Imaging Program, Division of Intramural Research, National Institute for Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH), Bethesda, Maryland, USA
Objectives: the antitubercular drug development process requires the use of animals mimicking the complex pathophysiology occurring in humans [1]. Several non-human primates (NHPs) have been used as models of human tuberculosis (TB) as they are more physiologically and immunologically similar to humans compared to other in vivo models of TB, and develop the full range of human-like lesions. The marmoset model of TB was developed specifically to support the drug development process due to the small size, ease of handling, low space and cost compared to other NHPs, and human-like pharmacology [1]. Nevertheless, the implications of pharmacokinetic (PK) differences have not been considered. While small, the marmoset is large enough for intra-individual serial blood sampling to determine the PK profiles at individual level [2]. Due to high inter-individual variability (IIV), a previous model-based analysis highlighted the need for a better characterization of drug disposition via a PK study [3].
This work aimed at developing PK models of the standard of care drugs rifampicin (RIF), isoniazid (INH), pyrazinamide (PZA) and ethambutol (EMB) tested in the aforementioned PK study in marmosets, integrating historical data. This step will allow us to assess the feasibility of using NHPs to support dose selection of novel anti-TB drugs.
Methods: in study 1 (historical data) plasma concentration-time profiles were obtained after single oral (PO) administration in uninfected marmosets with a rich sampling scheme, and after administration of multiple PO doses (RIF 15 mg/kg, EMB 50 mg/kg, PZA 125 mg/kg, INH 30 mg/kg) in infected animals with a sparse sampling scheme. In study 2 (new experiment), blood concentration data were obtained after the administration of the standard of care drugs in uninfected marmosets by intravenous (IV) route the first day, and PO the following days (RIF 15 mg/kg, EMB 50 mg/kg, PZA 75 mg/kg, INH 30 mg/kg), using a rich sampling scheme. This was done in order to characterize the oral bioavailability (F), except for PZA for which an IV formulation was not available. In both experiments the PK sampling after the first day was conducted at steady state. As a first step, the data of study 2 was modelled separately in order to estimate F whenever possible as IV and PO administration was performed in the same subjects. Subsequently, the data of both studies were modelled together, fixing F to the value identified in the previous step. In case of RIF, the introduction of clearance (CL) autoinduction in the model was considered, while for PZA a non-linear elimination was evaluated during the model building process. The analysis was done with NONMEM v7.5. The model assessment was performed through the evaluation of the objective function, parameter estimates precision, goodness of fit plots and visual predictive checks.
Results: a two compartment model structure with linear absorption and elimination was selected for EMB, the estimated CL was 0.236 L/h (100% IIV). A one compartment model with linear absorption and elimination was selected for INH and RIF. The estimated CL of INH was 0.261 L/h (40% IIV). For RIF, autoinduction of CL was observed and quantified. The non-induced estimated CL was 0.0592 L/h (50% IIV), which increased by 80% by the end of the study. For PZA a one compartment model with non-linear elimination with saturation was modelled with estimated Vmax of 11 mg/h (30% IIV) and kM of 100 mg/L. The estimated F was 0.5 for EMB, 0.4 for INH 0.6 and 0.4 for RIF. Inter-study differences in exposure compatible with change in F was observed for EMB and PZA.
Conclusions: PK models were able to adequately describe the data and the analysis highlighted high inter-individual variability, and inter-study differences with regards to F. Our findings suggest a challenge in the interpretation and evaluation of the pharmacokinetic-pharmacodynamic relationships in anti-TB drugs, unless careful PK sampling is performed at individual level. This requirement, implies further experimental and ethical considerations. Without individual PK information, estimates of drug potency and efficacy may be biased.
Acknowledgments:
This work has received support from the Innovative Medicines Initiatives 2 Joint Undertaking (grant No 853989).
References:
[1] H. J. Yang, D. Wang, X. Wen, D. M. Weiner, and L. E. Via, “One Size Fits All? Not in In Vivo Modeling of Tuberculosis Chemotherapeutics,” Front. Cell. Infect. Microbiol., vol. 11, no. March, pp. 1–23, 2021, doi: 10.3389/fcimb.2021.613149.
[2] L. E. Via et al., “A sterilizing tuberculosis treatment regimen is associated with faster clearance of bacteria in cavitary lesions in marmosets,” Antimicrob. Agents Chemother., vol. 59, no. 7, pp. 4181–4189, 2015, doi: 10.1128/AAC.00115-15.
[3] S. Grandoni et al., “Protocol optimization for the evaluation of pharmacokinetics, biomarkers and efficacy of antitubercular drugs in a novel tuberculosis infection model in marmosets,” 2023, [Online]. Available: https://www.page-meeting.org/default.asp?abstract=10684.
Reference: PAGE 32 (2024) Abstr 11042 [www.page-meeting.org/?abstract=11042]
Poster: Drug/Disease Modelling - Infection