Model-based experimental design recommendations for in vivo studies of the novel anti-tuberculosis drug MPL-447 improve characterization of PK and PD
Lina Keutzer (1), Fadel Sayes (2), Alexandre Pawlik (2), Wafa Zouaoui-Frigui (2), Samira Boarbi (3), Vanessa Mathys (3), Pierre-Yves Adnet (3), Natalya Serbina (4), Roland Brosch (2), Santiago Ferrer Bazaga (5), Ulrika S.H. Simonsson (1), on behalf of the ERA4TB Consortium (6)
(1) Uppsala University, Sweden, (2) Institut Pasteur, France, (3) Sciensano, Belgium, (4) TB Alliance, USA, (5) Universidad Carlos III de Madrid, Spain, (6) era4tb.org
Introduction: The novel mycobacterial membrane protein large 3 (MmpL3) transporter inhibitor MPL-447 is being developed for tuberculosis (TB) treatment by TB Alliance. MPL-447 is the prodrug of the active moiety MPL-204. In vivo MPL-447 activity was evaluated in a series of studies through ERA4TB (European Regimen Accelerator for Tuberculosis) platform.
Two in vivo dose ranging studies in mice were planned to assess MPL-204 pharmacodynamics (PD), i.e. decrease in bacterial burden measured in CFU. Further aims were to characterize the pharmacokinetics (PK) in infected mice and to evaluate the pharmacokinetic-pharmacodynamic (PKPD) relationship. To maximize the information gained from the experiments while respecting the 3Rs (Replacement, Reduction and Refinement), the originally planned traditional experimental design was optimized in collaboration between experimentalists, pharmacometricians, asset owner and scientific advisors.
Objectives:
- Optimize the design of the in vivo PK and PD studies to maximize the information and to facilitate longitudinal analysis while retaining the 3Rs
- Evaluate the performance of the traditional versus novel study design
Methods: Two studies, one in a chronic and one in an acute TB mouse model were planned. For both studies, BALB/c mice were infected with the Mycobacterium tuberculosis Erdman strain and dosed with MPL-447 via oral gavage. The originally planned study design for PK and PD evaluation was adapted in collaboration between experimentalists, pharmacometricians, asset owner and scientific advisors in the ERA4TB consortium, to ensure a balance between facilitating longitudinal analysis and practical aspects.
Based on the PK and PD data obtained from the two studies, a MPL-204 PK and a semi-mechanistic PKPD model describing change in CFU over time were developed in NONMEM (1). Using the PK and PKPD model, the PsN (2) function stochastic simulation and estimation (SSE) was applied to compare accuracy and precision in PK and PKPD parameters using the traditional versus novel design.
Results: The novel design comprised of the same total number of mice as the traditional design, thus respecting the 3Rs. However, the total number of mice to be culled for PD were distributed across more timepoints to enable longitudinal analysis, thereby decreasing the sample size to 3 mice per group. For the chronic study, mice were planned to be culled for CFU determination at day 1, 10, 28, 35, 42, 49, 56 and 84 since study start. For the acute study, the selected timepoints for PD were day 1, 3, 10, 14, 17, 21, 24 and 31 since study start.
To characterize MPL-204 PK, the Zipper design (3) was implemented, allowing for longitudinal data collection while keeping the sampling burden for the animals low. Each animal was sampled from the tail vein twice, but with the Zipper design 7 sampling timepoints and 5 dose groups could be covered, allowing full characterization of MPL-204 PK in infected mice. 60 mice were sampled, resulting in a sample size of 3 mice per timepoint. Due to little variability in BALB/c mice and integrated, model-based analysis of the data, the sample size was deemed sufficient.
Based on the SSE, the imprecision [relative root mean squared error (rRMSE)] and relative bias in PK parameters estimated based on the novel design was generally lower compared to the traditional design. The rRMSE for the novel (traditional) design was 14.1% (12.3%), 20.0% (34.6%) and 8.5% (32.4%) for Vmax, km and Vc, respectively. The relative bias for the novel (traditional) design was 4.0% (3.8%), 6.6% (18.4%) and -2.9 % (-17.9%) for Vmax, km and Vc.
For parameters characterizing the PKPD relationship, the rRMSE for the novel (traditional) design was 1.4% (3.8%) and 9.0% (37.0%) for Emax and EC50, respectively. The relative bias for the novel (traditional) design was 0.3% (1.8%) and 2.0% (18.9%) for Emax and EC50, respectively.
Conclusion: The joint collaborative effort was key to the success of these in vivo experiments, considering practical restrictions, 3Rs as well as ensuring to have an informative study design. The novel design enabled characterization of the full PK profile in infected mice, efficacy estimates over a wide dose range, longitudinal efficacy characterization, and estimation of MPL-204 PK and PKPD parameters with higher precision and accuracy.
Funding: This work has received support from the Innovative Medicines Initiatives 2 Joint Undertaking (grant No 853989). http://www.imi.europa.eu
References:
[1] Beal S, Sheiner L, Boeckmann A, Bauer R. NONMEM 7.4 Users Guides [Internet]. Gaithersburg, MD.: ICON plc; 1989. Available from: https://nonmem.iconplc.com/nonmem743/guides
[2] Keizer RJ, Karlsson MO, Hooker A. Modeling and Simulation Workbench for NONMEM: Tutorial on Pirana, PsN, and Xpose. CPT Pharmacometrics Syst Pharmacol. 2013 Jun 26;2:e50.
[3] Chen C, Ortega F, Alameda L, Ferrer S, Simonsson USH. Population pharmacokinetics, optimised design and sample size determination for rifampicin, isoniazid, ethambutol and pyrazinamide in the mouse. European Journal of Pharmaceutical Sciences. 2016 Oct 10;93:319–33.