Svenja Kemmer1, Nick Corr1, Roberto Abbiati1, Lena Friberg2, Mohammed Cherkaoui1
1Roche Pharma Research and Early Development, 2Department of Pharmacy, Uppsala University
Introduction: Myelosuppression is a dose-limiting adverse event frequently associated with chemotherapeutic drugs, often leading to treatment interruption or discontinuation. Current preclinical approaches, including animal models, recurrently lack clinical translatability and raise ethical concerns [1,2]. Although currently available in vitro and in silico models provide valuable mechanistic insights, they generally remain limited to single hematopoietic lineages [3,4]. To address this gap, we propose the development of a translational quantitative systems pharmacology (QSP) model to predict drug-induced anemia, neutropenia, and thrombocytopenia based on in vitro readouts of an immunocompetent multilineage hematotoxicity assay. Objectives: We are following a translational two-step approach: The initial focus is to quantitatively characterize drug-induced pharmacodynamic (PD) effects on the hematopoiesis process based on an in vitro QSP model. In a second step, this model will be extended to the clinical context by integrating pharmacokinetics (PK) and relevant in vivo regulatory mechanisms, including potential feedback pathways. This abstract mainly focuses on the first aspect of the work. Methods: To characterize the hematopoiesis process and assess the pharmacodynamic (PD) effects of drugs, we used an in-house developed immunocompetent multilineage hematotoxicity assay [5]. The assay was conducted using primary mononuclear bone marrow cells from four healthy donors. Flow cytometry data of cellular readouts, among them erythrocytes, neutrophils, platelets and their progenitors, were generated across five concentrations per compound, covering clinically relevant exposure levels. To capture the dynamic response to drug treatments, measurements were taken at five time points over a ten-day period. A previously published in vitro QSP model [6] was adapted to integrate this data from the updated assay, which includes refined progenitor cell populations and dynamics in an autologous immunocompetent context. The model was extended to incorporate drug effects at the progenitor cell level for a set of small molecule compounds including the marketed compounds docetaxel and olaparib and additional internal compounds known to induce anemia, neutropenia, and/or thrombocytopenia. Negative control compounds without hematotoxic effects were included for reference. Model refinement was performed iteratively, incorporating literature knowledge, visual inspection of model fits, and parameter sensitivity analysis. Parameter estimation was conducted using maximum likelihood estimation, leveraging the in vitro assay data to quantify both baseline hematopoiesis and drug-induced perturbations. The model development was performed with the R package dMod [7]. Results: The novel in vitro QSP model is encoded as a system of 20 ordinary differential equations (ODEs), capturing 16 key stages of hematopoiesis and drug-induced effects on progenitor cell populations. The model successfully reproduces multilineage hematopoiesis dynamics over the ten-day period, including normal maturation processes, and captures pharmacologically induced suppression of the erythroid, neutrophil, and platelet lineages. Drug effects are concentration-dependent and vary between compounds, allowing for a compound-specific characterization of hematotoxicity profiles including EC50 and Emax estimates. Additionally, immune cell activation was assessed to exclude immune-mediated confounding effects, but no significant activation was observed for the analyzed compounds, so this aspect was not explicitly incorporated into the model. Conclusions: This in vitro QSP model serves as the basis for translational predictions of clinical hematotoxicity. The next step involves integrating the model with PK data and additional in vivo processes, such as cytokine-mediated feedback regulation via erythropoietin (EPO), thrombopoietin (TPO), and granulocyte-colony stimulating factor (GCSF), which play a critical role in hematopoiesis recovery. Patient blood count data, including red blood cells, neutrophils, platelets and hemoglobin levels, are used to calibrate these additional in vivo processes, increasing the physiological relevance of the model. This integrative approach aims to provide a novel in vitro–in silico framework for early hematotoxicity risk assessment, enabling more accurate predictions of myelosuppression and supporting the optimization of therapeutic strategies in drug development.
[1] Wendler and Wehling (2010) The translatability of animal models for clinical development: biomarkers and disease models. Curr. Opin. Pharmacol. 10(5): 601-606. [2] Worp et al. (2010) Can animal models of disease reliably inform human studies? PLOS Medicine. 7(3): 1549-1676. [3] Friberg et al. (2002) Model of chemotherapy-induced myelosuppression with parameter consistency across drugs. Journal of Clinical Oncology. 20(24): 4713–21. [4] Craig (2017) Towards quantitative systems pharmacology models of chemotherapy-induced neutropenia. CPT Pharmacometrics Syst. Pharmacol. 6(5): 293–304. [5] Corr N. (2022) Application of a throughput amenable immunocompetent microphysiologic hematotoxicity model to drug development. ALTEX Proceedings MPS World Summit. 10(1): 24. [6] Wilson et al. (2020) An in vitro quantitative systems pharmacology approach for deconvolving mechanisms of drug-induced, multilineage cytopenias. PLoS Comput Biol. 16(7): e1007620. [7] Kaschek et al. (2019) Dynamic modeling, parameter estimation, and uncertainty analysis in R. J Stat Softw. 88(10): 1-32
Reference: PAGE 33 (2025) Abstr 11498 [www.page-meeting.org/?abstract=11498]
Poster: Methodology - New Modelling Approaches