Causal pathways from cognitive ability to Alzheimer's disease

Abstract

Alzheimer’s disease (AD) is a debilitating neurodegenerative disorder with a
preclinical period of up to twenty years before its diagnosis. Although the
clinicopathological signatures of AD are well-established, the causal mechanisms
leading to a higher risk of AD are poorly understood.
Several factors impede the understanding of causal factors for AD, such as bias due
to confounding by effects of concomitant diseases and ageing, as well as reverse
causality. Mendelian randomization (MR) has increasingly been used to understand
causal relationships, using genetic variants assigned at conception as instrumental
variables for the exposure, with minimal effects of confounding and reverse
causation, provided that the instrumental variable assumptions are fulfilled.
In my thesis, I aimed to identify aetiological factors increasing risk of AD by
employing bidirectional MR with age-stratified analyses. In Chapter 4 of my thesis, I
performed an age-stratified phenome-wide association study of the polygenic risk of
AD in UK Biobank to identify the earliest manifestations of AD risk. Phenotypes
associated with genetic liability to AD were followed up using MR to establish
whether they were likely to be causal risk factors. In Chapter 5 and Chapter 6, I
focused on brain morphology as a potential mediator of the association between
cognitive ability and AD, by performing bidirectional MR of both AD and cognitive
ability on structural brain measures. Where cognitive ability and AD were exposures
in the MR studies, these were performed in cohorts across the life course using the
ABCD study, Generation R, IMAGEN wave 1 for childhood, ALSPAC and IMAGEN
wave 2 for early adulthood, and UK Biobank (age-stratified) for later adulthood.
In Chapter 4, a wide range of phenotypes were associated with genetic liability to AD,
with cognitive ability being one of the phenotypes that was manifest in the youngest
tertile of UK Biobank, including and excluding the apolipoprotein E gene. In the
reverse direction, I found very little evidence of causal effects of these phenotypes on
AD risk, suggesting they are more likely reflecting the AD prodrome or survival bias.
In Chapter 5, I identified an age-dependent effect of genetic liability to AD on brain
morphology, but little evidence that brain morphology influences AD risk, providing
little support for the ‘brain reserve’ hypothesis. In Chapter 6, I found bidirectional
relationships between cognitive ability and regional and global structural brain
measures, with the largest effects observed with total surface area and estimated
total intracranial volume in both directions. The effects were consistent across the life
course, suggesting that genes for cognitive ability may operate through
neurodevelopmental mechanisms, highlighting childhood as a critical window for any
interventions. In conclusion, my findings suggest that structural brain morphology is
unlikely to be a mediator in the relationship between cognitive ability and AD.

Date of Award	21 Mar 2023
Original language	English
Awarding Institution	University of Bristol
Supervisor	Neil M Davies (Supervisor), Emma L Anderson (Supervisor), Laura D Howe (Supervisor) & Yoav Ben-Shlomo (Supervisor)