Modes of covariation between vascular risk factors and cognition in cohorts with cultural and geographical variations

Introduction

The number of people living with dementia worldwide is growing. Devising and implementing preventive strategies that are globally applicable is of paramount importance. However, previous studies have largely been limited to cohorts that are geographically and culturally homogenous. It is not known whether the associations between demographics, vascular risk factors (VRFs), and cognitive changes are consistent across cohorts with cultural and geographical variations.

Methods

Data were analysed retrospectively from 6 community-dwelling cohorts (> 60 years old) with cultural and geographical variations (British = 536; Hong Kong Chinese = 494; Australian = 302; Singaporean Chinese = 108; German = 102; Swedish = 94). Clinical demographics, VRFs, longitudinal cognitive changes, and MRI data (including T1, FLAIR, and diffusion images) were analysed from all 6 cohorts. First, across all cohorts, a grand canonical correlation analysis (n=1636) was used to establish common modes of covariation between demographics and VRFs on one side, and cognitive changes on the other side. Second, separately for each cohort, mediation analysis was used to investigate the mediating role of normalised total grey matter volume, microstructural integrity, white matter hyperintensity, and hippocampal structural network in the association mentioned above. Then the mediation outputs will be used in the meta- analysis to assess the indirect effect estimates and variance of each significant brain mediator.

Results

We identified three modes of covariation between dementia risk and cognition across all cohorts (p< 0.01). The strongest mode linked younger age and higher levels of education with better cognition at baseline and follow-up (Figure 1). Meta-analysis showed that, across cohorts, normalised total grey matter volume (Indirect effect = 0.02, Z = 3.13, p < 0.001, I² = 0%) was a significant mediator between demographics, VRFs, and cognitive decline after Bonferroni correction (Figure 2). The hippocampal structural network did not survive Bonferroni correlation while other brain mediators were not significant in the meta-analyses.

Discussion

This study investigated the complex relationship between demographics, VRFs, brain health, and cognition. Cohorts with cultural and demographic variations shared a common relationship between age, education, and cognition. However, individual cohort differences are detected in the contribution of different neuroimaging metrics.