(T-103) A model-based simulation workflow enables automated and accurate generation of clinical pharmacology summary statistics, a workflow and case-study.

Philip Delff, PhD – Director, Modeling and Simulation, Vertex Pharmaceuticals

Objectives: Noncompartmental analysis (NCA) is a useful method to calculate pharmacokinetic (PK) parameters, however, limitations due to study design, execution, or unexpected PK, can lead to inaccurate or biased results. A potential solution to this problem is to use a population PK model to simulate trial designs that can fill in the gaps to calculate the PK parameters. Here, we demonstrate a model-based simulation workflow for calculating PK parameters from single- and multiple-dose simulations and demonstrate a scenario where the multiple-dosing PK parameters accumulation ratio (AR) and effective half-life (EHL) can be inaccurately calculated by traditional methods, but accurately calculated via simulation-based methods.

Methods: Model-based simulations used NONMEM via the R package NMsim. The simulations were done using a published population PK model evaluating the effect of acid reducers on the absorption of dacomitinib. To calculate PK parameters, a single-dose simulation from time zero to infinity (30 days), and a multiple dose simulation with dosing to steady-state with observations from steady-state to 30 days after final dose, were run. Simulations were used to calculate and compare PK parameters to published NCA-derived parameters. The AR was calculated with commonly used equations as well as a less common equation (AUCinf,ss/AUCinf,single) enabled by this simulation approach.

Results: All code to apply this method are made publicly available. The parameters Tmax, Cmax, AUCtau, AUCinf, AR, and EHL were calculated after a single dose and at steady-state. The calculated PK parameters were close to the originally reported NCA parameters after a single 45mg dose of dacomitinib (Cmax: 12.0 vs 17.8 (50%CV), Tmax: 20h vs 8h (6-24h range), AUCinf: 966 vs 1234 (36%CV)). The AR was not reported in the original study but was calculated from simulations using Cmax, AUCtau, and AUCinf to be 3.52, 5.41, and 4.33, respectively, corresponding to EHL of 49.7, 81.4, and 63.3 hours. The mean dacomitinib half-life reported in the clinal study was 67 hours, indicating that the AR calculated using AUCinf,ss/AUCinf,single was more accurate than the traditional equations using Cmax or AUCtau. This method was only enabled through simulation, as AUCinf,ss and AUCinf,single are not typically measured in the same subject across MAD and SAD trials.

Conclusions: Population PK model-based simulations can be a powerful and, in some cases, essential tool to calculate accurate PK parameters of novel drugs. The publicly available simulation-based workflow presented can serve as a resource for researchers in need of a method to calculate PK parameters via simulation due to data limitations in PK studies. It was demonstrated that the AR and EHL can be more accurately calculated using AUCinf rather than AUCtau for drugs with a long absorption half-life, as was shown for dacomitinib, and this method is usually only possible with a simulation-based approach.

Citations: [1] Ruiz-Garcia, A., et al. Effect of Proton Pump inhibitor treatment on the bioavailability of dacomitinib in healthy volunteers. The Journal of Clinical Pharmacology. 2016 56: 223-230.
[2] Ruiz-Garcia A, et al. Pharmacokinetic Models to Characterize the Absorption Phase and the Influence of a Proton Pump Inhibitor on the Overall Exposure of Dacomitinib. Pharmaceutics. 2020 Apr 7;12(4):330.