Glioma tumours are among the most lethal brain disorders, claiming the lives of thousands of people in the United Kingdom each year. Despite the severity and prevalence of the condition, remarkably little is understood about the origins of gliomas, or the mechanisms that guide their spread within the brain. The aim of this thesis is to invoke a relatively new approach – brain network mapping – to provide insights into the origins of gliomas and their pathological spread along neural circuits. First, I provide a historical overview of both brain network mapping and glioma neurobiology, along with the recent advances and techniques popular in each field. In particular, I highlight preclinical research implying that gliomas originate from neural stem cells in the subventricular zone, as well as other work in mouse models demonstrating that gliomas infiltrate previously healthy brain networks. This thesis contributes three studies of clinical datasets which evaluate the hypothesis that glioma initiation and progression are guided by brain networks.
In the first study, I describe convergent evidence from both intracranial electrocorticography recordings and resting-state functional imaging of four patients with low-grade gliomas that tumour-infiltrated cortex can participate in large-scale cognitive circuits responsive to executive function. These findings imply that gliomas integrate into neural circuits, suggesting that their development and maintenance could be sustained by functional brain networks. In support of this idea, I next demonstrate that the spatial distribution of gliomas in the brain follows the distribution of functional network hubs, as well as cellular and genomic factors related to gliomagenesis. These results suggest two possibilities regarding the origins of glioma: the predilection of gliomas to hub locations could be a result of the vulnerability of hubs to oncogenesis, or the result of tumours arriving at central network locations while spreading through brain networks. To help disambiguate between these possibilities, I developed a novel approach termed “lesion covariance network mapping” to identify networks of brain regions co-lesioned in glioma, which indicate areas along which tumours are inferred to spread. This method revealed that gliomas cluster around horns of the lateral ventricles, consistent with the hypothesis that these tumours originate from neurogenic niches within subventricular zone. The lesion covariance network method also demonstrated that glioma localisation patterns follow specific structural and functional connectivity networks disseminating from periventricular grey matter. Cumulatively, the findings of the thesis support a model wherein periventricular brain connectivity guides glioma development from the subventricular zone into distributed regions of the cortex. In the conclusion, I discuss potential clinical applications of the presented research, such as in supporting predictive modelling approaches to forecast glioma progression, for the purpose of planning pre-emptive radiation and surgical treatments of glioma.