Hub overload and failure as a final common pathway in neurological brain network disorders

Abstract

Understanding the concept of network hubs and their role in brain disease is now rapidly becoming important for clinical neurology. Hub nodes in brain networks are areas highly connected to the rest of the brain, which handle a large part of all the network traffic. They also show high levels of neural activity and metabolism, which makes them vulnerable to many different types of pathology. The present review examines recent evidence for the prevalence and nature of hub involvement in a variety of neurological disorders, emphasizing common themes across different types of pathology. In focal epilepsy, pathological hubs may play a role in spreading of seizure activity, and removal of such hub nodes is associated with improved outcome. In stroke, damage to hubs is associated with impaired cognitive recovery. Breakdown of optimal brain network organization in multiple sclerosis is accompanied by cognitive dysfunction. In Alzheimer's disease, hyperactive hub nodes are directly associated with amyloid-beta and tau pathology. Early and reliable detection of hub pathology and disturbed connectivity in Alzheimer's disease with imaging and neurophysiological techniques opens up opportunities to detect patients with a network hyperexcitability profile, who could benefit from treatment with anti-epileptic drugs.

Keywords: Alzheimer’s disease; Cascading failure; Epilepsy; Hubs; Multiple sclerosis; Stroke.

Figure 1.

Possible role of pathological hubs in focal epilepsy. The primary epileptic focus, indicated in orange, is a local brain area with an abnormally high excitability. This can show up in EEG or MEG recordings as focal interictal epileptiform discharges (spikes) or high-frequency oscillations (HFOs). Often such an epileptic focus is surrounded by a zone of inhibition, indicated by the blue circle. If the local inhibition breaks down, abnormal activity from the focus can spread to a nearby pathological hubs. This is a region that has developed abnormally high structural and functional connectivity due to damage and partial recovery. When abnormal activation reaches the pathological hub, it may spread to other regions, including the highly connected physiological hubs such as the precuneus and posterior cingulum. From there, abnormal activation can rapidly spread to the rest of the brain, causing a generalized seizure. This schema explains how pathological hubs are different from the primary focus, have a distinct neurophysiological signature (hyperconnectivity instead of epileptiform discharges), and constitute an important relay station in the process of seizure generalization. Removal of pathological hubs may improve outcome in epilepsy surgery.

Figure 2.

Hub overload and failure. Schematic illustration of proposed hub overload and failure scenario: (A) Under normal conditions the brain network has many nodes with relatively low degree and a few nodes with higher degrees such as the nodes indicated by the green arrows (degree 4) and the red arrow (degree 5). Traffic on the network (action potentials propagated along axons connecting spatially separate brain regions) is distributed proportionally to degree, so the more highly connected hub nodes handle more of the traffic than the lower degree nodes. (B) When part of the brain is damaged, such as the area indicated in red, the traffic is redistributed over the remaining healthy nodes. For instance, the length two path between the nodes indicated by the blue and the green arrow no longer exists, and is replaced by a length three path, which in this case involves the highly connected degree 5 hub node. (C) Traffic redistribution is proportional to node degree. The highest degree node with degree 5 shows the greatest increase in activity (firing rates of neurons in this brain region) and functional connectivity to its neighbors, indicated by the red colors. The slightly lower degree 4 nodes show a smaller increase in activity. (D) When the extra load on the hub is too high and/or persists too long, this hub will be damaged. As a consequence, its activity and functional connectivity will decrease, and network traffic is now redirected to the next hubs in line indicated by yellow. (E) Now, these new, pathological hubs will show, in turn, an increased level of activity and hyperconnectivity to their neighbors. (F) If these new hubs are overloaded and fail, they will also end up with low levels of activity and connectivity. In this way, damage can cascade through the whole network, severely damaging its organization and function.