Multiscale network neuroscience in neuro-oncology: How tumors, brain networks, and behavior connect across scales

Abstract

Network neuroscience refers to the investigation of brain networks across different spatial and temporal scales, and has become a leading framework to understand the biology and functioning of the brain. In neuro-oncology, the study of brain networks has revealed many insights into the structure and function of cells, circuits, and the entire brain, and their association with both functional status (e.g., cognition) and survival. This review connects network findings from different scales of investigation, with the combined aim of informing neuro-oncological healthcare professionals on this exciting new field and also delineating the promising avenues for future translational and clinical research that may allow for application of network methods in neuro-oncological care.

Keywords: connectome; functional connectivity; glioma; graph theory; structural connectivity.

Figure 1.

Networks and their most important features. In (A), different types of networks consisting of 5 nodes (eg, node_i and node_j) are indicated. The left network has binary (present or not) and undirected (bidirectional) links. The middle network has binary links as well, but these are directed as indicated by the arrow heads. The right network has weighted links, with the colormap indicating that bright yellow links have a high weight, and darker purple links have low weight. The thickness of the lines also represents the weight in this figure. (B) depicts the matrix representation of the weighted network at the top right with the same color scale. Each row and each column represent the nodes in the network, while each element captures the weight of the link between pairs of nodes. The diagonal of the matrix, that is, the connection between a node and itself, is drawn in black to reflect that self-loops are not considered in this network. Moreover, other elements are black if a connection is not present according to the left-most network of panel A. (C) Schematically reflects integration between 2 exemplar orange nodes, as indicated by the dotted orange line with a path length of 2. The clustering coefficient of the purple node is calculated by dividing the total number of connections present between the neighbors (one solid purple line in this case) by the total number of possible connections between a node’s neighbors (3 in this case, as indicated by all purple lines, also the dotted nonpresent links), which yields a value of 0.33. In (D), the hubness (i.e., total summed weight of a node’s connections which reflects node strength) of each node is indicated through its color code.

Figure 2.

Schematic figure on network analysis of the brain across scales. Brain networks can be assessed in (A) humans at the macroscale using imaging techniques (structural, diffusion, and functional MRI), or neurophysiological recordings (magnetoencephalography or electroencephalography), in (B) rodent models using MRI, EEG, MEA, in vivo, or ex vivo calcium imaging or electrophysiology recordings, or in (C) brain cell cultures using MEA, calcium imaging recordings or electrophysiology.

Figure 3.

Multiscale network findings in neuro-oncology. (A) depicts clinically relevant outcomes in glioma patients. (B) reflects the types of global brain network abnormalities observed in glioma patients. In (C), a hyperactive neuronal network surrounding the tumor is displayed. The heightened neuronal activity leads to more secretion of glutamate around and within the tumor. Neurons in orange, tumor cells in blue, glutamate as orange circles and cellular activity in pink. In (D), neurons are shown to form synapses onto tumor cells where glutamate secretion activates tumor cells. The activation of one tumor cell gets propagated to neighboring tumor cells via tumor microtubes. The activation of tumor cells leads to glutamate excretion further enhancing tumor activity. Across all panels, arrows are drawn to indicate that findings from different scales may go hand in hand potentially due to causal relationships.

Figure 4.

Schematic figure depicting potential future clinical applications of multiscale network neuroscience along the disease course. Brain network analysis studies might aid clinical practice at multiple stages ranging from prognostication, resective strategy, disease-modifying treatment, disease and treatment monitoring to symptom management.