Salome Dumasdelage1, Blaise Pasquiers1
1Translational Pharmacometrics, Servier
Objectives: An antibody-drug conjugate (ADC) is under development for the treatment of cancer. Pharmacokinetic (PK) and tumor growth inhibition (TGI) studies have been conducted in xenograft mice models. This study is a first effort to estimate active exposure through a simple modelling approach and focused on characterizing PK and PK/PD relationship to extrapolate the dose-efficacy relationship and support dose selection in clinical setting through translational population PKPD modeling. Materials and Methods: The ADC was administered to a humanized-FcRn (huFcRn) mouse model at a dose of 10 mg/kg by intravenous (IV) route. In this study, PK samples were collected in 8 mice at 4 different times, with 2 mice per time point. For the PK/PD relationship, ADC was administered by IV, with doses ranging from 0.5 mg/kg to 2 mg/kg. Different administration schedules were tested, ranging from once a week to once a month. PK and TGI data were collected in 66 mice xenografted with tumor cells, including 6 controls for TGI. The translational approach was divided into PK and PK/PD relationship. PK was modeled and translated to human using allometry-based approach from huFcRn mice. Subsequently, the PK/PD model of the ADC was developped in xenografted mice, and several resistance mechanisms were tested. Since the reduction in tumor size was not immediate an exponential growth model with delayed treatment effect was used. Transit compartments were tested to represent the progressive mechanism leading to cell death [1]. For the resistance, innate resistance was modeled as pre-existing resistant cells, while acquired resistance emerged from treatment effects on sensitive or damaged cells. Assumptions included varying proliferation rates and treatment influences over time. Different models (linear, power, Emax) assessed treatment impact on each cell types. Model validation was conducted using goodness-of-fit (GOF) plots and visual predictive checks (VPC) for each dose level and administration schedule. Finally, these models were used to perform simulations in human to inform clinical translation. Monolix and SimulX (Lixoft) software were used for all modeling and simulation steps. Results: PK in huFcRn mice was successfully described using a two-compartment model, with linear elimination, with an estimated clearance of 0.23 mL/h/kg in huFcRn mice. For the PKPD relationship, PK and PD (TGI) were evaluated sequentially in xenografted mice. PK in this mouse model was well described by a two-compartment model with linear elimination. Tumor growth followed an exponential pattern. A saturation of the ADC effect at high doses was observed, which was described by an Emax model. The tumor growth rate (kge) was estimated at 0.0067 h-1. The tumor baseline value (TS0) was estimated at 166.67 mm3. The mean transit time (MTT) was estimated at 40 h. Additionally, tumor resistance to ADC was observed, manifesting as a loss of drug efficacy over time. To account for this phenomenon, a resistance mechanism was incorporated into the model, successfully capturing the diminishing treatment effect. For this model, all population parameters were well-estimated, including inter-individual variability on kge and TS0. The combined residual error model included a proportional component of 28% and an additive component of 6.95 mm³. Finally, PK/PD model-based simulations guided dose selection and optimized the administration regimen for clinical translation, demonstrating that efficacy could be maintained without requiring an intensive dosing schedule. Conclusion: Translational PK model providing human clearance predictions and PK/PD modeling successfully characterized ADC efficacy, capturing both the delayed tumor response and resistance mechanisms. Model-based simulations have guided first dosing strategies for clinical applications. This approach underscores the value of PKPD modeling and simulation to optimize ADC dosing strategies and to facilitate their translation into clinical development. Some mechanistic components such as payload release and target-mediated drug disposition (TMDD) still needs to be assessed for definitive extrapolation to human.
[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.
Reference: PAGE 33 (2025) Abstr 11415 [www.page-meeting.org/?abstract=11415]
Poster: Drug/Disease Modelling - Oncology