Cancer metastases are responsible for more than 90% of solid tumor deaths. Metastatic tumors are typically non-homogeneous and consist of diversified cellular clones with different proliferation and invasion abilities and different motility modes. We use computer simulations based on experimental data to study the role of diversity and heterogeneity in metastatic dissemination. The tissue and extracellular matrix are represented by a maze-like geometry, which captures physical constraints as well as signaling components. We focus on both the single cell level as well as collective and cooperative effects in a colony of cells.

The dissemination of different phenotypes with high and low proliferation and invasion rates is tested under different metabolic conditions. We find that each clone is optimized for different conditions, and their coexistence acts as a “bet hedging” strategy, for efficient coping with a rapidly changing environment. We also look at possible cooperative effects between cells with mesenchymal and amoeboid motility, moving together in a dense environment. The mesenchymal cells are able to degrade the blocking extracellular matrix (“path generators”), while the amoeboid cells are unable to degrade it but exhibit a more random trajectory (“path finders”). We show that a small population of mesenchymal cells can enhance the invasion of the entire, mostly amoeboid population. Direct cell-cell interaction, as evident for example in hybrid epithelial-mesenchymal cells, is found to be advantageous under metabolic stress but detrimental under normal conditions. We therefore predict that cell-cell interaction should be regulated as a stress-response mechanism. Our results are in agreement with experimental results of several cell types.