The syndromes associated with frontotemporal dementia are heterogeneous in their presentation and progression, with variable correlation between clinical phenotype and underlying proteinopathy. Single pathologies are associated with diverse clinical presentations, while the same clinical presentation can be caused by multiple pathological entities. Heterogeneity makes predicting underlying pathology and longitudinal outcomes challenging in clinical practice and in research settings. I propose that a multi-modal imaging approach, including structural and task-free functional magnetic resonance imaging, will provide mechanistic insight into how phenotypic variance arises and improve predictions of disease progression and survival.

In this thesis I draw from data for participants recruited at the University of Cambridge and from two multi-site collaborations, the Progressive Supranuclear Palsy Corticobasal Syndrome Multiple System Atrophy Longitudinal Study UK (PROSPECT-M-UK) and the Genetic Frontotemporal Dementia Initiative (GENFI). I describe characteristic differences in markers derived from task-free functional MRI and their relationship to patients’ clinical manifestations. I relate these functional changes to imaging markers of neuronal loss, cell death and synaptic loss. I find that subcortical atrophy from structural MRI relates to cortical functional network disruption, and that synaptic loss measured through [¹¹C]UCB-J positron emission tomography affects behaviour in relation to changes in functional connectivity.

I investigate differences in functional connectivity across the disease course. In individuals with familial frontotemporal dementia, time-varying functional network abnormalities predict symptomatic conversion in presymptomatic mutation carriers and future cognitive decline in symptomatic participants. In progressive supranuclear palsy and corticobasal syndrome between-network connectivity explains variability in survival but does not improve predictive accuracy beyond clinical and structural imaging metrics.

Imaging-derived biomarkers in frontotemporal lobar degeneration need to be appropriately targeted at components of the neurodegenerative cascade. Task-free functional MRI is an objective and scalable neural marker of clinical syndrome, useful in detecting symptomatic onset and prognostication but limited by small effect sizes, poor signal-to-noise ratio, and moderate reliability. I discuss developments required in image acquisition and analysis to support clinical practice and trials of experimental treatments.