Early identification of Alzheimer's disease: dynamic biomarkers for enrichment of trials

Alzheimer's disease (AD) is a fatal disease which in the UK alone directly affects 820,000 people, costing the UK economy £23 billion pounds, more than cancer and heart disease combined. All current treatments for Alzheimer's disease (AD) only provide relief from symptoms, despite many attempts to develop a cure. Recent research has shown that brain scans can be used to spot early signs of AD, up to 20 years before a clinical diagnosis. Researchers are now racing to develop drugs that can delay the onset of AD, based on the belief that it will be easier to delay rather than reverse the disease process. The major barrier in the development of such a treatment is the expensive brain scans necessary to identify people with early signs of AD. This fellowship will develop a relatively inexpensive alternative to make it practical to identify large numbers of people with the early signs of AD, in whom drugs to delay AD onset will be tested.

Within the fellowship Dr. Steven Kiddle will need to apply and develop statistical approaches to integrate complex data from brain scans, genetics, functional genomics and cognitive tests. He will be mentored by an expert in biostatistics, Dr. Chris Wallace from Cambridge University, and will collaborate with clinically trained researchers Professor Simon Lovestone and Dr. Claire Steves. Past research by Dr. Steven Kiddle and others have identified blood proteins which could be measured in a blood test, this could identify early signs of AD as an alternative to expensive brain scans. However, these early findings need much further validation before being used in the clinic. During this fellowship Dr. Kiddle will examine the potential of these blood proteins, as well as genetics and scores from cognitive tests, to identify subjects with early signs of AD.

This work will begin with a study in twins to determine which blood protein levels are affected by their current health, controlling for genetics. This will include a study that uses both blood tests and brain scans in identical twins, to see if the proposed blood test can reveal differences in early signs of AD in genetically identical individuals. Further work will attempt to identify early signs of AD using cognitive tests, by adapting existing assessments of cognitive abilities. Cognitive tests consist of a set of individual tasks which test different aspects of memory and cognition. Total scores for existing cognitive tests are known to be sensitive to late AD, but are thought to be less sensitive to early stage AD. However, individual tasks, or a clever combination of tasks, have not yet been assessed for ability to identify people in the early stages of AD. In addition, the combination of genetic, cognitive and blood markers may allow subjects to be identified with greater accuracy than each marker alone.

For blood markers to reflect signs of AD in the brain they must pass the blood-brain barrier. To investigate this I will study post-mortem brains from people who had been diagnosed with AD in life, and an equal number of people who had not recieved an AD diagnosis. Integrative analysis will be performed to link data on neuropathology, genetics, and multiple genomic levels, such as: epigenetics (which regulates gene expression), gene expression and protein levels. Dr. Kiddle will collaborate with Dr. Chris Wallace at Cambridge University to further develop statistical approaches tailored to this approach.

Finally, Dr. Kiddle will use statistical models to study causal relationships between early signs of AD and the best genetic, blood protein and cognitive markers. This novel approach will reveal which blood protein and cognitive markers dynamically reflect the early signs of AD, and therefore will reveal which markers should be studied in the future. Funding will be sought to setup large-scale studies to assess the utility of these markers.

Aim: To use integrated biostatistics and bioinformatics to relate genetic, peripheral and cognitive biomarkers to central Alzhiemer's disease (AD) pathology, in order to reduce barriers to clinical trials in the prodromal phase of the disease.

Objectives

1. Characterise genetic and environmental influences on candidate protein markers.
2. Identify tasks from cognitive tests that reflect amyloid pathology.
3. Relate markers to central pathology using a systems biology approach applied to post-mortem brains.
4. Build a causal model of genetic factors, brain Abeta, cognitive test items and candidate protein markers.

Methods
Objective 1
Sample: 106 twin pairs (for main analysis)
Analyses: An established quantitative-genetic model to test for genetic and environmental control of plasma protein levels and cognitive ability. Analysis of association of protein levels with other AD-related phenotypes, and with genetic data.

Objective 2
Sample: 144 + 273 + 56 = 473 subjects combined (across cohorts)
Analyses: Meta-analysis of association of cognitive task scores with brain amyloid beta levels. Multivariate regression and machine learning to find optimal combination of cognitive task scores.

Objective 3
Sample: ~200 case and ~200 control brains (for main analysis)
Analyses: Candidate markers will be assessed for differential expression in AD brains, taking into account genetic differences that affect their levels. Common genetic control of these traits will be further explored using 'co-localisation' analysis. Finally, genome-wide analyses on multiple genomic levels will be performed in an integrated fashion.

Objective 4
Sample: 116 + 273 + 170 = 559 combined subjects (across cohorts)
Analyses: The most promising markers from Objectives 1-3 will be studied further, with the causal relationship between genetics, brain amyloid levels and these markers investigated using Structural Equation Modelling.

Who will benefit from this research?

This research has a potential to affect the lives who people who currently don't display symptoms of Alzheimer's disease (AD), but who are slowly developing amyloid plaques that will cause AD-related neuro-degeneration and eventually death. Currently 820,000 people in the UK have AD, and 25 million people in the UK have a close friend or family member with dementia, of which AD is the most common form. As asymptomatic AD pathology is believed to develop over 20 years into clinical AD, there must be a substantial number of asymptomatic people currently developing AD pathology. Currently no treatment exists to either cure AD or prevent it's onset. The general public would also benefit from improved treatment options for AD because of the substantial costs of AD to the UK economy, which is estimated to be £23 billion, more than cancer and heart disease combined.

This research also has the potential to benefit the pharmaceutical industry by reducing costs of AD prevention trials. This will allow industry to trial more of their promising treatments in subjects most likely to benefit from such treatment, to the mutual benefit of participants and industry. Early detection of cancer has proved critical in its successfull treatment, and the development of new treatment options for cancer, this is likely to eventually be the case for other diseases such as AD too.

Potential outcomes from this research:

(1) The development of a procedure to identify people at risk of developing AD. The minimally invasive, and relatively inexpensive procedure could be used to predict subjects at risk from AD, before clinical symptoms.

(2) Identification of novel AD mechanisms to be targeted by future treatments. Novel biological processes or mechanisms that can be targeted by future treatments will be revealed from brain and blood studies.

(3) Understanding of the potential utility of plasma proteins as biomarkers of disease. By discovering plasma proteins that dynamically reflect current health, biomarkers of many diseases can be identified.

How will they benefit from this research?

Academic: Outcomes (1) and (2) will benefit AD researchers in the short term. As well as allowing replication, outcome (2) will provide novel AD mechanisms that can be followed up by treatment developers (in academia and industry) working on cell and animal models of AD. Outcome (3) can be used by researchers studying plasma biomarkers of a wide range of different diseases, to prioritise the study of proteins that are most reflective of current health (rather than by the static genotype).

Industry: Outcome (1) will benefit industry by helping to select asymptomatic individuals with elevated brain amyloid - the earliest sign of the development of AD - for prevention trials. Subjects whose biomarkers indicated a risk of elevated brain amyloid could be re-screened, using sensitive and specific amyloid-PET scans, this could reduce the total amount of PET scans needed to recruit a given number of individuals with elevated brain amyloid.

Patients: Outcome (1) would allow the recruitment of more subjects onto prevention trials, giving patients a chance to try innovative treatments. In the long term outcomes (1) and (2) could separately, or in conjunction, lead to the development of an effective treatment for AD.

General public: My fellowship will allow me to engage with people affected by AD, and to the wider public. Outcomes (1) and (2) could lead to a treatment for AD, which would affect the lives of the 25 million people in the UK population who have a close friend or family member with dementia. This has the potential to reduce the £23 billion AD is believed to cost the UK yearly.