zum Inhalt springen


Neurodegeneration Forecast: A Computational Brainsphere Model for Simulation of Alzheimer´s Disease

Fig.1 PET Images of amyloid-plaque, tau tangles in brain in vivo and fibre-tracking_DTI and functional MRI obtained by NM-UoC group

Accumulation of intraneuronal “tau-tangles” is a hallmark of Alzheimer’s Disease (AD). Together with extraneuronal aggregation of beta-amyloid peptides, tau-tangles are believed to be an important factor in AD and other neurodegenerative disorders. The tau-protein is physiologically involved in the stabilization of intraneuronal microtubule. In AD, hyperphosphorylated tau proteins dissolve from the microtubules and aggregate in the neurons in form of tangles, affecting neuronal function and leading to neuronal death. Whereas the tau-tangles in the brain of AD patients has been known for more than a century, only recently information has been collected (Drzezga et al. 2011; Riedl et al. 2014) indicating that they may travel from one neuron to another, inducing tangles in neighboring neurons in a prion-like fashion. This novel data is in good correspondence with recent studies indicating that protein aggregation pathologies as well as neurodegeneration itself appears to affect the brain not homogenously but spreads within certain brain functional networks. The striking similarity of the distribution of neuropathology with the anatomy of brain functional networks has raised questions on the pathways and mechanisms of pathology across the brain. The network neurodegeneration hypothesis postulates that patterns of neurodegeneration may be a result of the way the disease is moving within brain networks.

Gaining further evidence on tau-pathology in AD along brain networks is essential to better understand disease pathomechanisms and to predict the future course of disease. In addition, this may allow the development of novel biomarkers for monitoring disease modifying therapies. In this context, modern neuroimaging procedures represent a unique new option to gain insights in the distribution of neurodegenerative pathology in vivo and its relation to cerebral network architecture. Molecular imaging tracers for positron emission tomography (PET) today allow assessing the distribution of neuropathology such as tau tangles and amyloid-aggregation in the brain in vivo, noninvasively (Fig. 1). In addition, advanced magnetic resonance imaging methods (MRI) allow the assessment of parameters of network architecture in the brain. This includes fibre-tracking/DTI methods for assessing brain structural connectivity and resting state functional MRI methods for measuring brain functional connectivity (Fig. 1). The combination of the various imaging methods allow the unravelling of the interrelations between protein aggregation pathology and brain network structure. NM-UoC has longstanding expertise in cross-modal correlation between molecular imaging and functional/structural MRI imaging in neurodegenerative disorders, including studies on the network hypothesis.

A major hurdle in interpretation of this valuable in vivo data is the analysis of the complex interplay between multiple factors of pathology and the complexity of the brain networks. In mathematical-natural sciences, computational models have been developed to predict the spatial and temporal behavior of complex system. For example, Global Climate Models (GCMs) have been developed for weather forecast and climate projection. For modelling the development of a specific pathology in the brain networks, it is highly temping to develop a Global Brainsphere Model (GBM). Here, much can be learned from meteorological models, in philosophy, concept, physical principles, computational methods and pattern analysis techniques (Vereecken et al. 2016).

Model development for AD simulation is at its infant stage but gathering fast steam. Attempts have been made to use models for AD projections. For example, Raj et al. (2012; 2015) proposed a network diffusion model to simulate the trans-neuronal spread of AD pathology and predict future atrophy/metabolism states. However, in their study, the interactions of the neuron network and the brain physical-chemical environment are not considered. Hao and Friedman (2016) established a model of coupled equations for microscopic interactions of proteins. They emphasized on the model applications to clinical drug tests, but did not employ a brain network. Heaton et al. (2012) proposed a model for convective and diffusion on networks with exchanges with environment. Such models have, however, not yet been used in neurodegeneration studies. To our knowledge, a full-scale GBM has not been attempted, but will be tackled in this project.