Detecting, tracking & modelling structural and functional brain imaging changes in Alzheimer's disease

Dementia is a major and growing public health challenge and a research priority in the developed world. According to Alzheimer's Research UK, 25 million people in Britain have a family member or close friend with dementia, and the cost to our economy is estimated to be £23-billion per year.

There are currently only symptomatic treatments for Alzheimer's disease (AD); no 'disease-modifying' drugs. A lot of current research work relates to the search for these much needed disease-modifying treatments. Improved methods for assessing disease progression would increase the chances of showing that an intervention slowed the course of decline.

Structural changes, measurable on brain scans (magnetic resonance imaging; MRI) have been shown to predate and predict symptom onset and correlate with clinical decline in AD, and may help to evaluate potential disease-modifying therapies. Scans which appear visually normal to an expert might have subtle pathology detectable with computational analysis. There is emerging evidence that functional brain imaging (fMRI) can reveal changes even earlier in the course of the disease, and that these changes are associated with clinical measures like memory performance. There is a strong need for more precise characterisation of both structural and functional change, and for better understanding of the interactions between structure and function.

I will work to address these challenging problems in my Fellowship by developing new methods at the Wellcome Trust Centre for Neuroimaging, and applying these methods to large patient data-sets in collaboration with the Dementia Research Centre. Both Centres are part of the world-renowned UCL Institute of Neurology, at University College London.

My first objective is to develop a new statistical model for longitudinal structural imaging data (multiple brain scans over time), and to apply this to the problem of diagnosis and tracking of AD. This will allow hypotheses about the regional localisation of atrophy and its acceleration to be tested with greater power (i.e. we should be able to find changes using the new model that are undetectable without it).

My second aim is to investigate methods for modelling the connectivity among different brain regions, using fMRI acquired with subjects 'at rest'. Such data, and the brain networks that it can reveal, are generating increasing interest in the scientific and clinical research communities. Even in the very early stages of dementia, there may be changes in the way one brain region communicates with another. Reductions in these changes with a potential drug therapy could indicate success of that drug much earlier than clinical symptoms like memory loss. The utility of such 'biomarkers' from fMRI for dementia is currently an under-researched topic with great potential for detecting early changes and providing new outcome measures for treatment trials.

Finally, I aim to use the new methods together to investigate the inter-relation of structure and function and their changes in the disease. This will allow me to evaluate the power of the novel structural, functional and combined multi-modal measures for tracking disease progression. This has significant potential to provide better outcome measures for clinical trials, in terms of detecting change more robustly, less variably, or earlier in the disease course, with consequent impact on the search for disease-modifying treatments.

This proposal aims to assess cerebral structural and functional changes in Alzheimer's disease (AD). I will develop models for structural and functional changes and structure-function relationships using magnetic resonance imaging (MRI). Novel modelling approaches have the potential to yield scientific insights into the disease process and to provide improved biomarkers with impact on clinical studies and trials of candidate disease-modifying therapies. Specifically, I first propose to enhance the widely used statistical parametric mapping framework with a new spatio-temporal Bayesian model for serial MRI data, which will enable efficient modelling of the trajectories of volume change at every voxel in the brain, allowing direct localisation of regions with significant group differences or associations with clinical measurements, in terms of local volumes, rates of change of volume, and acceleration or deceleration. Secondly, I will investigate the application to AD of a recently developed method for Bayesian estimation and comparison of dynamical system models of brain connectivity (dynamic causal modelling; DCM) using functional MRI acquired at rest. Finally, I will explore the inter-relations of structural and functional changes, developing and evaluating biomarkers that combine their unique strengths. The methods will be widely applicable to imaging of dementia, but are particularly motivated by data from the Dominantly Inherited Alzheimer Network, available at the Dementia Research Centre, which includes longitudinal structural MRI and resting-state functional MRI in subjects at risk of familial AD, including presymptomatic mutation-carriers.

I have three hypotheses: that improved modelling of longitudinal structural data will result in more accurate and less variable estimation of disease progress; that measures of brain connectivity can provide biomarkers that change earlier in the disease process; and that combining functional and structural information will provide better biomarkers. If one or more of these hypotheses can be shown to hold, then this would have substantial impact on the search for disease-modifying therapies. The earlier the changes can be picked up, and the more accurately we are able to track progression and evaluate progress-slowing drug-effects, the greater the chances of finding successful treatments.

Improved methods related to diagnosis and tracking of progression have the potential for benefits to patient care in terms of earlier and more accurate diagnosis and prognosis. Methods for more sensitive measurement of treatment effects have the potential for short-term impact on pharmaceutical companies and other organisations involved in clinical trials of candidate disease-modifying therapies. In the longer term, these improvements for evaluating potential treatments, could contribute to the discovery of a successful treatment, which would then have very substantial impacts on patients, their carers, public health and the economy, due to reduced direct and indirect costs of care.