Advancing Neuroimaging-Based Brain Network Models in Neurodegenerative Diseases

Abstract

In this thesis I contribute to the broader scientific effort of advancing neuroimaging-based brain network models for neurodegenerative diseases, with a particular focus on Alzheimer's disease (AD). First, I provide novel insights on the multifactorial and heterogeneous pathophysiology of AD by investigating the role of co-occurring pathologies (e.g., cerebrovascular disease, CSVD). Second, I introduce these concepts into brain network models. By doing so, I advance our current understanding of the impact of co-pathologies on functional brain activity, providing information needed to build more biologically realistic brain network models. <br/> In the first project, I demonstrate that alterations in perivascular spaces - an emerging biomarker of CSVD – are observed in young individuals with a genetic form of AD. This finding suggests that CSVD is an intrinsic feature of AD rather than a coincidental consequence of aging. In the second project, I show in a cohort of late-onset AD the crucial role that white matter hyperintensities – a key biomarker of CSVD - have on cortical neurodegeneration. These insights confirm the role of CSVD in AD, beyond traditional AD biomarkers, such as amyloid and tau. In the third project, using a brain network model, I demonstrate the detrimental effect of white matter hyperintensities on global brain network communication in wakeful resting-state. In the fourth project, I show that a reduction in inter-hemispheric white matter connectivity, which is commonly observed in aging, AD, and CSVD, explains brain alterations observed during slow-wave sleep. <br/> By providing novel insights into the complex and multifactorial nature of AD pathophysiology and by introducing these newly developed concepts into brain network models, this thesis effectively advances their biological plausibility. This step is needed to increase the overall reliability of these models, which will help to translate them from the laptop to the bedside. An effective clinical translation of brain network models could potentially provide new tools to improve diagnosis, prognosis and treatment personalization in neurodegenerative diseases.