The metastasis of malignant epithelial tumors begins with the egress of transformed cells from the confines of their basement membrane to their surrounding collagenous stroma. The invasion can be morphologically diverse, ranging from that of dispersed mesenchymal cells to multicellular collectives. When cancer cells are cultured within basement membrane-like matrix (BM), or Type 1 collagen, or a combination of both, they show collective-, dispersed mesenchymal-, and hybrid collective-dispersed (multiscale) invasion respectively. Here, we ask how distinct these invasive modes are with respect to the cellular and microenvironmental cues that drive them, and how can cancer cells transition between such states. A rigorous exploration of invasion was performed within an experimentally motivated Cellular Potts-based computational modeling environment. The model comprises of adhesive interactions between cancer cells, BM- and collagen-like extracellular matrix (ECM), and reaction-diffusion-based remodeling of ECM. The model output consisted of metrics cognate to dispersed- and collective- invasion. An exhaustive combination of input values gave rise to a spatial output distribution that comprised dispersed-, collective- and multiscale- invasion. K-means clustering of the output distribution followed by silhouette analysis revealed three optimal clusters: one signifying indolent invasion and two representing multiscale invasions, each of which encompass the purer dispersed- and collective invasions respectively. Principal component analyses of phenotypic outputs close to the separation of these two multiscale modes established specific input signatures: perturbing such signatures revealed the cues that can allow phenotypic transitions among these three clusters. Our systems-level analysis provides quantitative insights into the biases and constraints cancer cells face during their invasion through tissue microenvironments with distinct physicochemical properties.### Competing Interest StatementThe authors have declared no competing interest.

Interactive dynamics of reaction-diffusion and adhesion predict diverse invasion strategies of cancer cells in matrix-like microenvironments