Establishing the Spatial and Cognitive Specificity of Cerebrovascular Burden: A UK Biobank Study

2176 

Abstract Submission 

Katie Moran¹, Nils Muhlert², Daniela Montaldi¹

¹University of Manchester, Manchester, Greater Manchester, ²University of Manchester, Manchester, UK

Katie Moran  
University of Manchester
Manchester, Greater Manchester

Nils Muhlert  
University of Manchester
Manchester, UK

Daniela Montaldi  
University of Manchester
Manchester, Greater Manchester

Cerebrovascular (CV) risk factors are shown to have a detrimental impact on cognitive function and are likely to do so through the disruption of key white matter (WM) networks. Identifying early neurocognitive indicators of poor CV health may be critical in preventing later-life cognitive dysfunction.
The heightened sensitivity of WM to CV insult is well established (Iadecola, 2013), but increasing evidence shows this effect may be region-specific within WM. For example, hypertension, a known CV risk factor, is associated with greater change in anterior CV function, such as reduced blood flow, relative to posterior regions (Kandil et al., 2022; Beason-Held et al., 2007). This shows the unique CV aetiological processes between regions and highlights an anterior-specific vulnerability to ill CV health, which is likely also reflected by anterior WM change. Given the proposed region-specific impact of CV burden, it's likely that the cognitive impact of this is also specific to certain cognitive domains. Processing speed (PS) is a likely candidate given its links to WM (Kerchner et al., 2012), CV health (Cox et al., 2019) and anterior degeneration (Kochunov et al., 2010). As such, our current work aims to establish the relationship between CV burden, WM and cognition and to assess the regional and cognitive specificity of this effect.

The sample of 35,000 from the UK Biobank contained both males and females aged between 45-70, with complete data. A baseline SEM was built whereby CV burden acted as the predictor variable, loaded with 7 known CV risk factors. A latent construct of anterior WM integrity was the mediator variable, loaded with averages of 7 anterior tracts with separate models made for FA & MD (figure 1a). We ran a PCA on cognitive data from the UK Biobank to derive two principal components. One which represented general cognitive performance and a second which separately captured PS. The PS principal component was used as the outcome variable for the baseline SEM. To establish spatial specificity, a new SEM was built, identical to the baseline model, but the anterior WM integrity was replaced with posterior WM integrity (Figure 1b) as a mediator. To establish cognitive specificity, a third SEM was built, here PS was replaced with cognitive performance as an outcome variable. Models were then statistically compared.

All measured variables loaded significantly onto each latent construct. The baseline SEM illustrated that increased CVB is associated with slowed cognitive PS. This is mediated by anterior WM integrity loss. All pathways in the baseline model were significant and the model fit well (Figure 2a). Replacing anterior WM integrity with posterior WM integrity either reduced the mediation effect (FA) or it became no longer significant (MD) (Figure 2b). The model also fit significantly worse, when compared with the baseline model (p < .001). When the outcome measure was replaced with general cognitive performance, the direct effect remained significant. However, the mediation either reduced (FA) or became no longer significant (MD) (Figure 2c). Model fit indices demonstrated good fit, with no significant change from baseline.

The findings demonstrate increased CV burden was associated with slowed PS. This effect was at least partly mediated by diffusion changes in anterior WM, which was shown to be stronger and more consistent than posterior WM, in line with previous work. While CV burden was associated with both cognitive measures, changes in anterior WM integrity best explained the relationship between CV burden and PS, as opposed to cognitive performance. The next empirical steps will focus on addressing why this regional vulnerability exists through the use of multimodal imaging techniques, establishing the timeline of CV-driven decline associated with each cognitive domain and finally, identifying the networks which underpin CV-driven cognitive decline in other domains using a graph theory approach.

Executive Function, Cognitive Control and Decision Making ²

Aging

Diffusion MRI Modeling and Analysis

White Matter Anatomy, Fiber Pathways and Connectivity ¹

Cerebrovascular Disease

Cognition

WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC

Other - Cerebrovascular