(T-040) How to Make a Salad? Rethinking Pharmacometric/QSP Model Composition Using Open-Source Julia Tools

Daniel Kirouac, PhD – Vice President of Translational and Systems Pharmacology, Metrum Research Group; Timothy Knab, PhD – Senior Scientist II, Metrum Research Group; Ellen Swanson, PhD – Research Scientist II, Metrum Research Group

Objectives: Pharmacometric and systems pharmacology models are often modular as different, independent components can be joined together to form a more complex model. The process of combining and reusing model components can be challenging with no clear framework and, as such, investigators often resort to rewriting models from scratch rather than reusing the individual components. Additionally, model components could be written in different notations such as ordinary differential equations (ODEs) or reactions, depending on the most convenient way to represent a system. This adds an additional complexity to the model composition process. A framework is presented that allows an investigator to seamlessly combine different model components represented in their respective notations and reuse these independent components to create multiple combinations of integrated models, just like mixing the components of a salad.

Methods: Julia [1] open-source tools, namely ModelingToolkit.jl [2] and Catalyst.jl [3], were used to present a convenient framework for pharmacometric model composition. The symbolic-numeric model representation of ModelingToolkit.jl and the reaction notation provided by Catalyst.jl allowed for seamless composition of independent model components presented as ODEs or reactions.

Results: The framework was demonstrated by composing different model components (e.g., Pharmacokinetic (PK), Pharmacodynamic (PD), physiological organs) to build larger models (e.g., PKPD, Physiologically Based PK (PBPK), Quantitative Systems Pharmacology (QSP)). Both ODEs and reaction notations were combined into integrated PKPD and QSP models with examples drawn from bispecific T cell engagers, viral dynamics, and drug-drug interactions (DDI). The framework enabled seamless transitions from in vitro to in vivo murine to clinical settings for a bispecific T cell engager application [4].

Conclusions: A framework based on Julia open-source tools was proposed in this work to allow for seamless pharmacometric and QSP model composition. This framework enables model reusability and translation using convenient and flexible model notation.

Citations: 1. Bezanson J, Edelman A, Karpinski S, Shah VB. Julia: A Fresh Approach to Numerical Computing. SIAM Rev. 2017;59: 65–98.
2. Ma Y, Gowda S, Anantharaman R, Laughman C, Shah V, Rackauckas C. ModelingToolkit: A Composable Graph Transformation System For Equation-Based Modeling. arXiv [cs.MS]. 2021. Available: https://arxiv.org/abs/2103.05244
3. Loman TE, Ma Y, Ilin V, Gowda S, Korsbo N, Yewale N, et al. Catalyst: Fast and flexible modeling of reaction networks. PLoS Comput Biol. 2023;19: e1011530.
4. Betts A, Haddish-Berhane N, Shah DK, van der Graaf PH, Barletta F, King L, et al. A translational quantitative systems pharmacology model for CD3 bispecific molecules: Application to quantify T cell-mediated tumor cell killing by P-cadherin LP DART®. AAPS J. 2019 May 22;21(4):66.