Blaise Pasquiers, Sylvain Fouliard
Translational Pharmacometrics, Servier, France
Objectives: A small molecule (drug S) is currently under development for treatment of cancer patients, targeting the regulation of apoptosis. In vivo pharmacokinetic (PK) and tumor growth inhibition (TGI) in xenograft mice have been generated. The objectives of this work were to extrapolate PK and PKPD relationship to human and explore the dose/efficacy relationship and the impact of dosing schedule using PBPK and popPKPD modeling.
Materials and Methods: In-vitro data (including molecular weight, logP, pKa, blood/plasma ratio, unbound fractions in plasma and liver, intrinsic clearance) for drug S were generated along with in vivo PK data in mice and TGI data in mice xenografted with tumor cells. A total of 80 mouse (including 13 controls) TGI profiles were collected. Drug S was administered via intravenous (IV) or subcutaneous (SC) routes, as single doses or various repeated dosing schedules with doses ranging from 7.5 to 15.
The translational approach was divided into PK and PK/PD relationships. PK was modeled using a physiologically based pharmacokinetic (PBPK) approach with Simcyp software. In vivo-in vitro extrapolation (IVIVE) of this model was measured in mice to assess its predictive capacity before human simulations.
The PKPD relationship of the product on TGI was analyzed using a population approach in Monolix. A linear/exponential growth model, with delayed treatment effect was used1. In order to describe a diminution of the effect over time, several resistance mechanisms were tested in this model. Innate resistance was implemented as a pool of resistant cells presents before the drug and acquired resistance was implemented as induced by treatment, directly from the sensitive pool cells or from one of the damaged cell compartments. Different assumptions regarding rate of proliferation (different, equal) were tested for sensitive and resistant cells, as well as time-dependence and treatment effect (linear, power, and Emax effects on sensitive and resistant cells). After validation, simulations of this model in human were conducted using PK predicted by the PBPK model. A dose-response relationship was described through multi-dose simulations.
Results: The PBPK model allowed good IVIVE in mice, with prediction-to-observation ratios of 1.93 and 0.83 for AUCinf and Cmax, respectively, with Cmax being the principal PK parameter as treatment effect was mainly driven by high concentrations. This model enabled the simulation of PK profiles of the molecule in human by adapting values related to clearance and distribution (blood/plasma ratio, unbound fractions in plasma and liver, intrinsic clearance).
The final PKPD model was developed fixing individual PK parameters to the values obtained in a previous linear two-compartment model. The PKPD model included an exponential/linear growth and 3 transit compartments after drug effect, with sensitive cells proliferating (Count at baseline of sensitive tumor cells (baseline_TS0): 226 mm3, exponential rate of proliferation (lambda0): 0.0088 day-1, linear rate of proliferation (lambda1): 7.67 mm3/day). By the effect of molecule S concentration, these cells undergo different evolutionary stages represented by compartments T1, T2, and T3 (first-order rate constant of transit, k1: 0.098 day-1) before dying. Acquired resistance was added to this model: T3 cells can die or start a new proliferation cycle as resistant cells, as described by Eigenmann et al.2. Their proliferation has the same rate as sensitive cells, and S product kills these resistant cells with a significantly reduced effect compared to sensitive cells. Consistently with an effect driven by highest concentrations, treatment effect was described by a power, with sensitive cells having ~150-fold lower sensitivity than resistant cells. Variability was described on baseline_TS0, lambda0, lambda1, and kkill_sensitive. Proportional error was estimated at 26%. All population parameters were well-estimated (RSE < 30%), and the model was validated through visual predictive checks for each dose, administration route, and administration schedules. This model was used to simulate early PKPD profiles in human to provide an early indication of effective dose.
Conclusion: PBPK and PKPD models, incorporating a resistance mechanism for molecule S in apoptosis, were successfully developed in mice, allowing early dose-efficacy relationship in human. This model is a first step in a model informed drug development approach in order to support early development of drug S.
References:
[1] Simeoni M, Magni P, Cammia C, De Nicolao G, Croci V, Pesenti E, Germani M, Poggesi I, Rocchetti M. Predictive pharmacokinetic-pharmacodynamic modeling of tumor growth kinetics in xenograft models after administration of anticancer agents. Cancer Res. 2004 Feb 1;64(3):1094-101. doi: 10.1158/0008-5472.can-03-2524. PMID: 14871843.
[2] Eigenmann MJ, Frances N, Lavé T, Walz AC. PKPD modeling of acquired resistance to anti-cancer drug treatment. J Pharmacokinet Pharmacodyn. 2017 Dec;44(6):617-630. doi: 10.1007/s10928-017-9553-x. Epub 2017 Oct 31. PMID: 29090407; PMCID: PMC5686279.
Reference: PAGE 32 (2024) Abstr 10820 [www.page-meeting.org/?abstract=10820]
Poster: Drug/Disease Modelling - Oncology