Temporal lobe connections and parallel networks for memory: possible mechanisms of resilience and recovery after stroke

Memory breakdown in older age is a major challenge for medical research, with an increasing burden in personal, societal and fiscal terms. Stroke is an important cause. A recent survey by the Stroke Association highlighted memory problems as one of the leading areas of unmet need in stroke survivors. Memory depends on widespread networks in the brain. These networks are connected by bundles of white matter, which essentially are the wiring of the brain. The connections are likely to be important. However, we know surprisingly little about the role of connections in memory or memory breakdown after stroke, despite the fact that minor strokes often damage white matter. Part of the reason is that until recently, it was not possible to identify and delineate the important connections for memory from brain images (e.g. MRI).

Our recent research used a technique called diffusion tensor MRI to tackle this challenge and investigate connections likely to be important for memory. A particular tract called the fornix was found to be most important in the healthy brain and in ageing. However, in individuals at an early stage of memory decline alternative pathways became disproportionately more important. This led to the idea that individuals with early memory decline might be especially vulnerable to injury to these alternative tracts. The purpose of this project will be to test this idea by investigating the effects of stroke. Study of the impact of stroke will also add to knowledge of how memory is organised in the normal brain.

The project will focus on two groups of individuals: a group of patients with recent stroke affecting white matter and a group of individuals from an ongoing study of healthy adults drawn from the population of Rotterdam, the Netherlands, the leading study of its kind in the world. Both groups will have MRI, including diffusion tensor MRI, and cognitive assessment. The importance of the second group lies in the fact that they have already been followed for a number of years, so information will be available on memory and cognitive trajectory before stroke, providing the strongest test of the idea of shifting vulnerability. The pattern of damage to temporal lobe connections in the brain will be assessed and related to the impact of stroke on memory. We will determine how undamaged tracts contribute to recovery over one year. Finally, cutting-edge computational image analysis techniques will be applied to try and predict memory profile in more detail and extract maximum information about prognosis from brain images. To date, this powerful technology has been applied to diagnosis but not to prognosis.

Understanding of the factors that dictate vulnerability, resilience or recovery of memory systems will aid the logical development of new treatments. Further, the ability to predict prognosis more accurately will help to target treatments to patients most likely to benefit and will improve the quality of information we can give to patients about how they are likely to recover.

The main objective is to study the effects of focal stroke affecting temporal lobe white matter tracts; and to use this knowledge to develop a scientifically informed framework to stratify patients based on accurate predictions of prognosis.

The design combines prospective investigation of two cohorts: a clinical cohort with recent stroke; and a cohort from an existing longitudinal population-based study with incident stroke. The combination of these cohorts is a key strength: the Rotterdam cohort offers replication, and the strongest test of the main hypothesis because of the presence of prospectively collected data on pre-stroke cognitive trajectory. The King's cohort will provide deeper insight into mechanisms, through more detailed characterisation of brain structure and memory profile. The potential for independent replication and cross-validation are key strengths.

Diffusion tensor MRI will be used to reconstruct and quantify the microstructural properties of major temporal association tracts implicated in memory, including fornix, uncinate and parahippocampal cingulum. The applicant's prior work has demonstrated the central role of the fornix in the healthy brain and in ageing, and that non-fornical tracts become disproportonately important in early memory breakdown. The diffusion and tractography methodology are established and validated in older and clinical groups. The functional relevance of specific tracts will be investigated further at the neuropsychological level, using tests that probe decision-making, semantic and familiarity aspects of memory.

Multivariate pattern recognition of images has been applied to diagnosis. Here, we will apply a Bayesian framework to the problem of predicting prognosis and stratification of patients based on prognosis. A novel aspect will be developing an approach to predict the whole profile of memory (rather than a single measure), and to try to extract additional prognostic information from brain images.

Economics and Society
The total economic costs of stroke to the UK in 2006/07 were £4.5 billion. Over half relates to health and social care costs. Cognitive impairment is a major contributor to this and to lost economic productivity in those who maintain independent living. Ameliorating this burden through clearer understanding, leading to better predictions of prognosis and better targeting of treatments to individuals is the ultimate aim of this proposal.

Communication and Engagement
Major non-academic beneficiaries of this project include patients, carers and the charities that support them. We have already found that ideas of connections and memory appeal to a broad range of interested groups and individuals. These concepts have a lot to contribute to the way memory is described to the public and the way that memory disorders are explained. Plans to communicate these perspectives include plans to engage with survivors of stroke, patients with early memory decline and healthy older adults.
- Stroke survivors. The findings will be described to stroke survivors through the applicant's links with the Stroke Association (as a member of the Research Advisory Committee). Dr O'Sullivan has written articles for the main magazine for patients and carers published by the Stroke Association and will prepare a piece on memory partly based on this research. The description of memory problems in the society's factsheets also needs to be ravamped in the light of recent research. The Stroke Association will be involved in the organisation of the healthcare professionals' symposium with the hope that this will lead to further engagement activities.
- Ageing. The current project will engage with local older adults through the public meetings of the Clinical Age Research Unit at King's College Hospital. This unit has had a number of public events since its inception, approximately annually but dependent on research activity, and this is a popular aspect of its work.
- Mild cognitive impairment and dementia. The Institute of Psychiatry already co-hosts an annual research day with the Alzheimer Society. This is highly successful, attracting 150 delegates in 2012 drawn from patients, carers, representatives of charities and local services and interested members of the community. The applicant is already disseminating his research findings through this forum.

Healthcare and Clinical Practice
The main practical application to emerge will be a model to predict memory outcome after stroke. An engagement event with healthcare professionals, including clinical psychologists, therapists and nurses, will provide the opportunity to shape day to day practice. The tools to implement advances based on automated pattern recognition of brain scans will be shared with the scientific community as open-source software. Rapid translation to clinical benefit is a principle of the Academic Health Sciences Centre at King's. A related example of how such an advance will be communicated and implemented is the use of machine-learning to support diagnosis of Alzheimer's disease, which has already been adopted in clinical services (see 'Early Dementia Diagnosis' on http://www.kcl.ac.uk/kingsanswers/news/). As in other areas of biomarker research, we will also exploit these results ourselves by the design of pragmatic studies (e.g. Health Technology Assessment).

Capacity
An important obstacle in raising public and professional awareness in cognitive aspects of stroke is lack of capacity and lack of both researchers and clinical staff specialising in this area. This project will begin to increase capacity, by providing a research focus of vascular cognitive disorders to several new researchers. Development of an interest and skills in engagement will be encouraged and fostered through links with the charities (Stroke Association and Alzheimer Society).