Contributed Talk

On the road to cellular digital twins of in vivo tumors

Julian Herold^1,2, Eric Behle³, Jakob Rosenbauer³, Jacopo Ferruzzi⁴, Alexander Schug³

Contact: 

To this day, cancer remains an insufficiently understood disease plaguing humanity. In particular, the mechanisms driving tumor invasion still require extensive study. Current investigations address collective cellular behavior within tumors, which leads to solid or fluid tissue dynamics. Furthermore, the extracellular matrix (ECM) has come into focus as a driving force facilitating invasion. To complement the experimental studies, computational models are employed, and advances in computational power within HPC systems have enabled the simulation of macroscopic tissue arrangements. We hereby present our work using Cells in Silico (CiS), a high performance framework for large-scale tissue simulation previously developed by us [1,2]. Combining a cellular Potts model and an agent-based layer, CiS is capable of simulating tissues composed of millions of cells, while accurately representing many physical and biological properties. Our ultimate aim is to build a cellular digital twin of an in vivo tumor. Unfortunately, current in vivo measurement methods lack the required resolution for directly parameterizing our simulations. Therefore, our current strategy is to parameterize CiS via a bottom-up approach, utilizing experimental data from multiple smaller in vitro systems. We focused our first studies on tumor spheroids, spherical aggregates composed of thousands of individual cells, which are one of the main workhorses of tumor analysis. The comparison between real and simulated spheroids represents a way to fully exploit both data sources, but remains a challenging task. Towards this, we developed a novel data-agnostic method to compare spatial features of spheroids in 3D [3]. To do so, we defined and extracted features from spheroid point cloud data, which we simulated using CiS. We then defined metrics to compare features between individual spheroids, and combined all metrics into an overall deviation score. Finally, we used our features to compare experimental data on invading spheroids in increasing collagen densities. We propose that our approach represents the basis for defining improved metrics to compare large 3D data sets. Moving forward, this approach will enable both basic and applied researchers to close the loop between modeling and experiments in cancer research.