Abstract
Agent-based models (ABMs) are readily used to capture the stochasticity in tumour evolution; however, these models are often challenging to validate with experimental measurements due to model complexity. The Voronoi cell-based model (VCBM) is an off-lattice agent-based model that captures individual cell shapes using a Voronoi tessellation and mimics the evolution of cancer cell proliferation and movement. Evidence suggests tumours can exhibit biphasic growth in vivo . To account for this phenomena, we extend the VCBM to capture the existence of two distinct growth phases. Prior work primarily focused on point estimation for the parameters without consideration of estimating uncertainty. In this paper, approximate Bayesian computation is employed to calibrate the model to in vivo measurements of breast, ovarian and pancreatic cancer. Our approach involves estimating the distribution of parameters that govern cancer cell proliferation and recovering outputs that match the experimental data. Our results show that the VCBM, and its biphasic extension, provides insight into tumour growth and quantifies uncertainty in the switching time between the two phases of the biphasic growth model. We find this approach enables precise estimates for the time taken for a daughter cell to become a mature cell. This allows us to propose future refinements to the model to improve accuracy, whilst also making conclusions about the differences in cancer cell characteristics.
Introduction
Tumour growth is a complex, stochastic process driven by individual cell interactions. While Agent-Based Models (ABMs) are well-suited to simulate this "bottom-up" behavior, they present significant calibration challenges. Because ABMs rely on random simulations rather than closed-form equations, they lack a standard likelihood function—the mathematical bridge typically required to connect a model to experimental data. Historically, researchers have relied on "best guess" parameters or simple point estimates. This study addresses the need for a rigorous statistical framework to calibrate the Voronoi Cell-Based Model (VCBM) and quantify parameter uncertainty using real-world measurements from breast, ovarian, and pancreatic cancers.
Methodology: The Voronoi Cell-Based Model
The research utilizes the Voronoi Cell-Based Model (VCBM), an off-lattice ABM where cells are represented by points and their physical boundaries are defined by Voronoi tessellation. Cell movement is governed by mechanical forces, and proliferation is limited by nutrient availability, modeled as the distance from the tumour boundary.
To calibrate this complex model, the authors employed Sequential Monte Carlo - Approximate Bayesian Computation (SMC-ABC). Since the model's likelihood is intractable, ABC involves simulating the model thousands of times with varying parameter sets and retaining only those that produce results close to observed data. The SMC approach enhances efficiency by iteratively narrowing the tolerance for acceptable matches. Furthermore, a "biphasic" extension was introduced to account for tumours that exhibit a sudden acceleration in growth rates, allowing parameters to shift at a specific time point.
Experimental Data and Calibration Results
The study calibrated the VCBM against three in vivo mouse-based datasets: breast cancer (66 days), ovarian cancer (25 days), and pancreatic cancer (33 days).
Key findings include:
- Ovarian Cancer: The standard monophasic (single-rate) VCBM provided an excellent fit for the steady growth observed in ovarian cancer data.
- Pancreatic Cancer: A single-phase model failed to capture the growth dynamics. However, the biphasic extension successfully modeled a dramatic "gear shift" where the cell cycle time ($g_{age}$) dropped significantly, leading to a growth spurt.
- Breast Cancer: The model predicted exponential growth, while experimental data trended linearly, suggesting the need for additional biological constraints like nutrient depletion or immune pressure in future iterations.
Discussion and Conclusion
The application of SMC-ABC provided full posterior distributions for parameters, offering a measure of confidence for biological inferences. While the model successfully identified growth shifts, it does not yet explain the underlying biological causes, such as genetic mutations. Future work will explore 3D modeling and "online" calibration to improve accuracy and computational efficiency. This framework demonstrates that complex ABMs can be rigorously validated, providing deeper insights into the mechanisms of tumour progression.
Summary updated