Perivascular spaces and their relationship with changes in cognition over time

Background

Perivascular spaces (PVS) are small fluid-filled gaps found around small blood vessels in the brain in the deep grey and white matter, and are involved in the removal of waste products. Impaired clearance of waste from the brain is associated with cognitive decline and has been linked to enlarged PVS. We have developed and improved image analysis methodology to measure PVS metrics such as volume and length using both T1 and T2 MRI scans and this project will investigate the role of PVS metrics in the prediction of clinically relevant outcomes such as future cognitive decline. 

We have already conducted meta-analyses with data from ~20,000 individuals and now wish to explore the impact of assumptions of PVS shape and clinical presentation on the accuracy and usefulness of statistical models using PVS to predict outcomes such as dementia. 

The project aim is to assess the role of PVS, in addition to known predictors such as age, in the development of cognitive decline and the potential for PVS in decision making with and planning for individual patients in the clinical setting. 

We have recently formed the International Perivascular Spaces Meta-analysis Collaboration (IPVSMC) with a number of international partners, which aims to foster availability of data and the technical skills required to derive and use PVS metrics. The student will both contribute to IPVSMC and work with collaborators across the world. 

This is an interdisciplinary project, requiring input from medical, statistical, and image analyst colleagues in a world-leading department in the field of brain imaging. The student should have a quantitative background in a subject such as statistics, biology, physics or similar and be passionate about using data to find solutions to real-world problems. 

Aims

We have PVS, clinical, and cognitive data from studies recruiting participants with different clinical presentations (for example, stroke, sleep apnoea, and large population studies including UK Biobank). This will allow for the exploration of PVS in groups with different baseline risks of cognitive decline and a variety of risk factors such as high blood pressure. The studies differ in a number of important ways, such as the tool used to measure cognition, and the student will investigate the impact of these differences on the analyses and the conclusions we may draw from them, with particular emphasis on the way the role of PVS can change and its relevance to longitudinal cognitive status. 

Training Outcomes

1. Understanding and use of statistical models, in particular with their application to meta-analysis of individual participant data. The training will also comprise the methodology of prediction modelling and use of discrimination and calibration statistics to assess and fine-tune model performance. There are also considerations re model complexity versus model usability. 
2. Skills in use of statistical software such as SAS and R. 
3. Gain understanding of key brain imaging appearances and their relevance to stroke and dementia, and of inter-relationships with vascular risk factors and cognitive status. 
4. Data management, use of meta-data, future-proofing of datasets to allow use beyond current planned analyses.
5. Data governance, particularly with respect to the legal framework required for sharing data both nationally and internationally.
6. Understanding of image analysis methodology and impact on statistical analyses. 

References

Valdés Hernández MDC, Duarte Coello R, Xu W, Bernal J, Cheng Y, Ballerini L, Wiseman SJ, Chappell FM, Clancy U, Jaime García D, Arteaga Reyes C, Zhang JF, Liu X, Hewins W, Stringer M, Doubal F, Thrippleton MJ, Jochems A, Brown R, Wardlaw JM. Influence of threshold selection and image sequence in in-vivo segmentation of enlarged perivascular spaces. J Neurosci Methods. 2024;403:110037. doi: 10.1016/j.jneumeth.2023.110037. 

Steyerberg EW, Vergouwe Y. Towards better clinical prediction models: seven steps for development and an ABCD for validation. Eur Heart J. 2014;35(29):1925-31. doi: 10.1093/eurheartj/ehu207. 

Romero JR, Pinheiro A, Aparicio HJ, DeCarli CS, Demissie S, Seshadri S. MRI-Visible Perivascular Spaces and Risk of Incident Dementia: The Framingham Heart Study. Neurology. 2022;99(23):e2561-e2571. doi: 10.1212/WNL.0000000000201293.