Collective dynamics and model verification: Connecting kinetic modeling to data

Towards understanding glioma growth patterns: is glioma migration stochastic or deterministic

Pedro Lowenstein

University of Michigan School of Medicine


At the clinical level it is not understood how brain tumors grow to large sizes without causing midline shifts or without causing serious symptoms. At the experimental level how brain tumors are built up from individual cells within a tissue has not been elucidated. To understand brain tumor pattern formation we performed a series of experimental, modeling, and eventually, clinical studies. To examine tumor formation from the earliest stages we followed the detailed paths taken by fluorescently labeled glioma cells in the host brain. Glioma cells became associated with particular microanatomical brain compartments. Indeed, there exist glioma cells from mouse, rat and human glioma tumors that become closely associated with brain vessels, and appear to grow solely on brain blood vessels. This close association implies that their growth pattern will disrupt oxygenation levels locally, as glioma migration patterns dislodge astrocyte end feet from the blood vessels. We predict that this growth pattern kills brain tissue situated between adjacent blood vessels colonized by glioma cells; this creates space that can occupied by dividing glioma cells. As this pattern is repeated iteratively, it leads to the formation of a macroscopic tumor by gradually replacing the host brain tissue. This could explain why brain tumors grow to large sizes without causing major symptoms or without displacing the brain's midline, until late in the disease. To test whether this simple mechanism could explain the formation of brain tumors by glioma cells that grow exclusively on blood vessels we constructed an agent based model. The model, which did not include angiogenesis, reproduced the iterative growth pattern detected in vivo. That glioma cells can be found associated with vessels, white matter tracts or interstitially has been known for more than 80 years. Our experiments beg the question of whether individual glioma cells grow exclusively on blood vessels, or on white matter tracts, or interstitially. In other words, is glioma migration and invasion stochastic or deterministic. If it is stochastic then glioma cells can become associated with all compartments; if it is deterministic then some glioma cells will only be associated with one single compartment. Based on the findings and model described above, we hypothesized that glioma migration is deterministic and that different cells have preferential migration pathways in the brain. We will provide preliminary evidence showing the existence of human glioma cells that are exclusively associated with one individual brain microanatomical compartment. Summarizing, our work shows how a fruitful dialectical interaction between experimental biologists and mathematicians can help discover new syntheses in our understanding of brain tumor biology.