Resting state fMRI as a marker for progression from mild cognitive impairment to Alzheimer’s disease

Sammanfattning: The arc of progression in the most common neurological aiction known as Alzheimer's Disease (AD), is characterized by a prodromal stage of Mild Cognitive Impairment (MCI). MCI subjects have traditionally been diagnosed with a battery of cognitive tests, but in recent times two good biomarker predictors of incipient AD have been identied. The cerebrospinal uid levels of the protein residues amyloid-beta and phosphorylated tau can be quantied and directly relate to the imprint of associated pathologies in the brain. This work aims to elucidate the impact of tau- and amyloid-related pathologies on the functional networks of the brain, as gauged by a resting-state fMRI connectivity analysis. In this context, we aim to identify optimal model parameters that yield maximal contrast between subspecies of MCI and healthy controls, such as the most sensitive frequency interval for the Blood Oxygenation Level Dependent (BOLD) time-series and the resolution of whole-brain parcellation schemes. The connectivity analysis exposes the impact of biomarker pathology and outlines a tentative progression pattern, relating the decline of functional connectivity to increasingly pathological levels of biomarkers. A progression hypothesisis proposed, reviewing pattern progression in the light of neuronal communication breakdown and phase-lag. Furthermore, failure of key hubs are identied using graph theoretical centrality measures and the relative group separations with connectivity pattern is evaluated by means of support-vector machines. Relative healthy controls, MCI with non-pathological CSF levels of biomarkers exhibit a widespread pattern of reduced connectivity, likely due to a mix many dementia subtypes. MCI subjects with pathological amyloid CSF levels but normal values of tau, has a large set of failing links converg- ing on crucial network hubs: thalamus, caudate nucleus and putamen, and are additionally aected in key regions such as hippocampus. This nding supports the view of Alzheimer's progression in terms of global disconnection syndrome by failing hub regions. Furthermore, these patterns are man- ifested in relevant graph theoretical centrality measures. MCI with pathological levels of both CSF biomarkers produce the strongest contrast relative healthy controls, involving reduced connectivity beteween parietal and frontal areas, but also implicating areas linked with cognitive decline, such as hippocampus and posterior cingulate cortex. The similarity of this contrast to that of controls vs. Alzheimer's subjects, indicates the presence of a functional progression pattern with two biomarker levels. Our ndings merit further investigation of the biomarker progression line using larger cohorts further stratied with cognitive test scores.