Justin Wilkins(1), Wilhelmina Bagchus (2), Claude Oeuvray (3), Akash Khandelwal(4)
(1) Occams, Amstelveen, The Netherlands, (2) Merck Institute for Pharmacometrics, Merck Serono SA, Lausanne, Switzerland, (3) Global Health Institute of Merck, Switzerland (4) Merck Healthcare KGaA, Darmstadt, Germany
Objectives: M5717 is a novel antimalarial under development for single-dose treatment, transmission blocking and chemoprotection (1). It has a long half-life and duration of action and is active against multiple life-cycle stages of Plasmodium falciparum and P. vivax.
The purpose of this analysis was to build population pharmacokinetic-pharmacodynamic (PK/PD) models to characterize dose-exposure and exposure-response relationships of M5717 in healthy volunteers in terms of parasite clearance following an induced blood stage malaria (IBSM) challenge.
This presentation will focus on how the complexity in this system was reduced to allow a practical predictive model to be developed.
Methods: In total, 1542 blood concentrations across 11 dose levels (50-2100 mg) from 71 subjects and 513 parasite concentrations across 3 dose levels (150 mg, 400 mg, 800 mg) from 22 subjects collected during a Phase I trial were available for analysis. There were 21.7 PK and 23.3 PD observations per subject.
NONMEM was used to build a nonlinear mixed-effects PK model for the population. Interindividual variability (IIV) was tested on all structural parameters. This was expanded into a PK/PD model incorporating the effect of drug concentration on parasitaemia in the IBSM population. Complex models mimicking the parasite life cycle and semi-physiological approaches to explaining an observed lag in drug effect on parasite clearance (attributable to the presence of non-viable parasites that, while “dead”, were still detectable) seemed feasible (2), along with simpler models defined in terms of parasite growth and clearance (3). The M3 method was used to compensate for the large number of PD concentrations below the quantification limit (32.5%).
Results: PK were complex, with highly variable absorption, repeated secondary peaks, and apparent dose-nonproportionality in volume. The final model for M5717 PK was three-compartmental, with first-order elimination, a transit absorption model in combination with first-order absorption, a recirculation model, and a strong inverse relationship between central volume of distribution (V2/F) and dose. Allometric weight was included a priori on clearance and volume parameters (4). Clearance (CL/F) was estimated to be 21.3 L/h (with an IIV of 24.8%), V2/F was estimated to be 2350 L (IIV 24.5%), peripheral volumes of distribution (V3/F, V4/F) were estimated to be 1540 L and 2450 L respectively, and intercompartmental clearances (Q2/F, Q3/F) were estimated to be 6.37 L/h and 58.5 L/h, respectively. Absorption rate (ka) was estimated to be 3.85 /h (IIV 117%), mean transit time from formulation to depot was estimated to be 0.414 h (IIV 65.9%), and the number of transit compartments was 5.57 (IIV 50.5%). Bioavailability (F) in the population was fixed to unity with an IIV of 24.1%. A power coefficient of -0.530 defined the effect of dose on V2/F.
Periodic oscillations in parasitaemia reflecting stages in the parasite’s life cycle were not accounted for, but this aspect was irrelevant to the model’s purpose and had no impact on the results. Baseline parasite count at the time of challenge was estimated to be 0.0358 parasites/mL (IIV 128%). Parasite growth rate in the absence of drug was estimated to be 0.0635 /h (IIV 12.0%), maximal parasite clearance rate was estimated to be 0.218 /h (IIV 20.9%), EC50 was estimated to be 8.20 ng/mL (IIV 50.5%), and the Hill coefficient for the relationship was 19.0. Delay in effect onset had a half-life of 26.3 h.
Model parameters were well estimated, with relatively high precision, despite a high signal-to-noise ratio. Standard and simulation-based diagnostics confirmed an adequate fit of the model to the data and its suitability for simulation.
The minimum inhibitory concentration (MIC) was calculated to be 7.59 ng/mL (95% CI 2.84-20.6 ng/mL). The parasiticidal concentration required for 90% killing (MPC90) was calculated to be 9.21 ng/mL (95% CI 3.45-25.0 ng/mL).
Conclusions: Sometimes even large amounts of densely-sampled data are insufficient to support the modelling of complex dynamic systems. Working within time constraints requires pragmatism – models must serve the purpose for which they are built, balancing “perfection” with practicality of development and application.
Despite the unusually intricate nature of the system, a (somewhat) parsimonious approach was appropriate for developing a predictive model for M5717 PK/PD, suitable for use in dose-finding simulations.
References:
[1] Baragaña B, Hallyburton I, Lee MCS, Norcross NR, Grimaldi R, Otto TD, et al. A novel multiple-stage antimalarial agent that inhibits protein synthesis. Nature. 2015;522(7556):315–20.
[2] Hietala SF, Mårtensson A, Ngasala B, Dahlström S, Lindegårdh N, Annerberg A, et al. Population pharmacokinetics and pharmacodynamics of artemether and lumefantrine during combination treatment in children with uncomplicated falciparum malaria in Tanzania. Antimicrob Agents Chemother. 2010;54(11):4780–8.
[3] Krause A, Dingemanse J, Mathis A, Marquart L, Möhrle JJ, McCarthy JS. Pharmacokinetic/pharmacodynamic modelling of the antimalarial effect of Actelion-451840 in an induced blood stage malaria study in healthy subjects. Br J Clin Pharmacol. 2016;412–21.
[4] Anderson BJ, Holford NHG. Mechanism-Based Concepts of Size and Maturity in Pharmacokinetics. Annu Rev Pharmacol Toxicol. 2008;48(1):303–32.
Reference: PAGE 29 (2021) Abstr 9588 [www.page-meeting.org/?abstract=9588]
Poster: Drug/Disease Modelling - Infection