[POS] Systems Biology

[POS-02] Graph-Theoretical Analysis for Identifying Influential Components in Quantitative Systems Pharmacology Model Structures

*Ryoya Sano¹, Kota Toshimoto¹, Hiroyuki Sayama¹, Yu Chen², Kazuo Hiekata², Masayo Oishi¹ (1. Astellas Pharma Inc (Japan), 2. The University of Tokyo (Japan))

Keywords:

Quantitative Systems Pharmacology,Pharmacometrics

A Quantitative Systems Pharmacology (QSP) model is a mathematical and computational model used to characterize biological systems, disease processes, and drug pharmacology. By employing a system of ordinary differential equations, a QSP model can simulate the dynamics of biological components. This capability enables the quantitative interpretation of animal experimental data and the exploration of hypothetical scenarios (“what-if” analysis) before conducting clinical trials. Consequently, QSP modeling holds significant potential for improving the success rate of Proof of Concept in drug development.Constructing a QSP model requires a comprehensive understanding of interactions among biological components, such as disease-related target molecules and biomarkers. One of major challenges in QSP implementation is defining these interactions, particularly when developing a model from scratch for diseases with limited quantitative data. In such cases, models are often built based on disease pathway maps, resulting in highly complex structures. Identifying key components that significantly influence model outputs is crucial for refining model structures and prioritizing experimental studies. Additionally, one valuable application of QSP modeling in drug discovery research is target identification, which aids in determining therapeutic targets within complex disease pathway. In this context, understanding the impact of individual components on key outputs (e.g., tumor size in cancer) would strongly support the identification and validation of therapeutic targets, especially in the early stages of drug discovery. However, quantitatively ranking each component’s impact remains a significant challenge. In this study, a method was developed to estimate components that influence the target of interest using only the model structure of a QSP model. The network of interactions within the QSP model was represented as a directed graph, and a graph-theoretical approach, which estimates how perturbations to specific components propagate to the target of interest based solely on the graph-structure. Application of this method to published QSP model structures such as immuno-oncology revealed several components whose influence on the target was not intuitively expected. To validate these findings, perturbations were introduced into actual QSP models, resulting in significant effects on the target of interest for a part of these components.In conclusion, this method enables the efficient identification of influential components solely based on the network structure of biological systems and disease processes. By accelerating QSP model construction and aiding in target identification, this approach has the potential to enhance drug discovery and development.