Posted Tue, Jan, 19,2016
This author interview is by Dr. Vangelis Sakkalis, of Foundation for Research & Technology-Hellas. Dr Sakkalis' full paper, The Importance of Neighborhood Scheme Selection in Agent-based Tumor Growth Modeling, is available for download in Cancer Informatics.
Please summarize for readers the content of your article.
Tumors are highly complex, dynamical biological systems spanning multiple scales from the molecular to cell and further to tissue level. Numerous mathematical models and research work focus on understanding the underlying mechanisms with the ultimate aim to improve prognosis and therapeutic outcome. In this work, we present a hybrid mathematical approach to model tumor invasion, based on a model initially proposed by Anderson. Hybrid models address the multiscale nature of tumors by combining both discrete and continuous variables. In particular, tumor cells are treated as individual entities and are modeled in a discrete way using cellular automata. It is assumed that cells lie on a regular grid and are allowed to move, proliferate and die as interact locally with their microenvironment following a set of rules. The tumor microenvironment is comprised of oxygen and the Extracellular Matrix (ECM), which both are modeled using continuous variables. This work shows that allowing different neighbors in the grid to approximate cellular movement, significantly different results can arise with respect to tumor morphology and the evolution of its populations. Although lattice anisotropies arising from regular lattices can be circumvented by using random and isotropic lattices as well as by using off-lattice models, we demonstrate in this work how a simple extension from von Neumann to Moore neighborhood in cell migration on a regular lattice can better approximate cellular motility by minimizing the artifacts arising from lattice anisotropies and how the neighborhood selection can significantly affect tumor growth and morphology. Nevertheless, an extensive exploration of various microenvironments and tumor behaviors with increasing complexity is required in order to better approximate real tumor patterns and reveal the conditions under which the differences between the movement approaches can become more/less evident.
How did you come to be involved in your area of study?
I first came across this really interesting area, on a preparation meeting for a European proposal (named ContraCancrum), about 7 or 8 years ago. The aim was to enhance existing tumor simulators well beyond the state-of-the-art, especially on the biochemical level (molecular dynamics), on the molecular level (detailed molecular networks) and on the cellular and upper biocomplexity levels (angiogenesis, embryology considerations, biomechanics, medical image analysis etc.). At that time, clinical application of the simulation outcome was far beyond reality, yet still promising. The years after, exactly that missing link from theoretical computational modeling to real clinical applications worldwide, was the focus of our research group and it still remains a challenging topic.
What was previously known about the topic of your article?
Implementing a simplified version of the hybrid tumor growth model on a regular grid, we observed that the emerged tumors had a diamond shape instead of the circular shape that was expected from the radial symmetry of the problem under study. This also raised questions about the effect of these artifacts on simulation outcomes regarding tumor evolution. Previous studies have shown that artifacts arising from regular grids can be circumvented by the use of random and isotropic lattices or by using off-lattice models (Drasdo, Advances in Complex Systems, Vol. 8, Nos. 2 & 3 (2005) 319-363). Keeping in mind the popularity and simplicity of cellular automata models on regular lattices, we explored in this work the effect of neighborhood selection on tumor morphology and growth and we observed that a simple extension from von Neumann to Moore neighborhood in cell migration is capable of minimizing the artifacts arising from lattice anisotropies.
How has your work in this area advanced understanding of the topic?
This work shows how a simple extension from von Neumann to Moore neighborhood can minimize the artifacts arising from lattice anisotropies and how the neighborhood selection can significantly affect tumor expansion and morphology. This means that misleading conclusions regarding tumor evolution might arise based on the choice of the neighborhood.
The effect of neighborhood selection on tumor evolution should be reevaluated and further explored in various conditions, microenvironments and tumor behaviors with increasing complexity in order to better approximate real tumor patterns that aid in tumor understanding, while provide valid predictions of clinical importance.
What do you regard as being the most important aspect of the results reported in the article?
The emergence of the invasive behavior throughout tumor evolution is highly associated with a fatal outcome. As cellular migration is critical for tumor growth and invasion, it stresses the importance to describe movement as adequately as possible. Considering that cellular automata models on a regular lattice remain very popular for the modeling of tumor invasion, this work shows how a simple extension from von Neumann to Moore neighborhood can minimize the artifacts arising from lattice anisotropies and how the neighborhood selection can significantly affect tumor evolution. As an example, our results show that the Moore-based movement resulted in tumors that expand faster and form tumor patterns that highly differ from the corresponding von Neumann-based ones, which are considerably more compact.
Posted in: Authors
My experience publishing in Human Parasitic Diseases was very positive. I was very satisfied with the rapid and high-quality review process and the constructive feedback. The comments from the reviewers allowed me to improve the paper significantly. I highly recommend that other researchers publish their papers in Libertas Academia Journals.