Adrien Tessier, Sylvain Fouliard and Marylore Chenel
Clinical Pharmacokinetics and Pharmacometrics division, Servier, France
Objectives: S48168/ARM210 is proposed orally for the treatment of Duchenne Muscular Dystrophy (DMD) in children from 5 years old. A phase I program was performed in adults and consisted of a single (SAD) and multiple ascending dose (MAD) study, a formulation bridging study and a DDI study. A model-based strategy was proposed to predict the S48168/ARM210 PK in children and support dose selection for studies in the DMD population. With this purpose in mind, a population PK model was developed using data in adults and was allometrically scaled to children.
Methods: S48168/ARM210 plasma concentrations in 61 adult male healthy volunteers (HV, all Caucasian and CYP2C8 extensive metabolizers) from the SAD, MAD and DDI studies were used to develop the adult model for doses from 20 to 240 mg. Plasma concentrations in 11 subjects at 20 mg from the formulation bridging study were used externally to assess the formulation effect (a previous NCA analysis showed a relative bioavailability equal to 1).
Model estimation was performed using Monolix version 2016R1 software and SAEM algorithm.
The effect of demographic covariates such as body weight was not tested. Allometric functions on total body weight were applied without test on apparent distribution and elimination parameters assuming pre-defined fixed exponent of 3/4 for clearance and 1 for volume parameters [1]. Because the drug is proposed in children from 5 years old, metabolism pathways are assumed mature and no maturation function was used.
The effect of Gemfibrozil (a strong CYP2C8 inhibitor) co-administration in the DDI study on S48168/ARM210 PK was investigated as a categorical covariate on the relative bioavailability (F) and CL/F.
Using published body weights according to age [2], doses selection for the study in children was performed. Using the scaled model, PK profiles were simulated to derive Cmax and AUC at steady-state. Doses were selected for different weight-bands to achieve in children the same exposures as observed in adults from the MAD study for the two highest dose levels tested and that were safe (60 and 120 mg).
Results: The complex and highly variable absorption was best described by a sequential first- and zero-order absorption process with a lag-time and high IIV and IOV on some parameters. The disposition was best described by a 2-compartment model with a linear elimination and low IIV and IOV. The residual error was modeled using a combined error model.
A dose effect was significant on F and the duration of the zero-order process with three and two doses categories respectively. The effect of co-administration with Gemfibrozil was significant on CL/F and associated to a decrease by 37% of the typical value. Allometric functions based on total body weight were then included with fixed exponent (on V1/F, V2/F, Q/F and CL/F).
Model evaluation showed that the final model described adequately the drug plasma PK in adult HV. Visual Predictive Checks and Posterior Predictive Checks of derived parameters (Cmax and AUC) showed that the model is qualified to simulate new data.
Using simulations performed with the scaled PK model in children, 6 weight-bands (from 15 to more than 55 kg) were identified in order to keep the exposure in children within the range of exposures found in adults. One dose was proposed for each weight-band, the highest corresponding to the dose given in adults. Thus two set of doses were proposed corresponding to the two highest and safe dose levels tested in adults in the MAD study.
Conclusions:
A population PK model developed through a combined analysis can describe the pharmacokinetics of S48168/ARM210 in plasma for adult male HV. This model was qualified through different metrics. Scaling allometrically the model to fit known weight classes for children can then be used to pick potential doses to match resulting safe exposure levels in children to those previous established in healthy adult males. Both Cmax and AUC parameters can be used with one of those parameters dominating the dosage selection.
The scaled population PK model will be used during the first study in the DMD population in order to support the dose adaptation during the dose escalation process.
References:
[1] Holford NH (1996). A size standard for pharmacokinetics. Clin Pharmacokinet. 30(5):329-32
[2] West NA, Yang ML, Weitzenkamp DA, Andrews J, Meaney FJ, Oleszek J, Miller LA, Matthews D, DiGuiseppi C (2013). Patterns of growth in ambulatory males with Duchenne muscular dystrophy. J Pediatr. 163(6):1759-1763.e1. doi: 10.1016/j.jpeds.2013.08.004. Epub 2013 Oct 6.
Reference: PAGE 27 (2018) Abstr 8582 [www.page-meeting.org/?abstract=8582]
Poster: Drug/Disease Modelling - Paediatrics