DOI: 10.1002/alz.079738 ISSN: 1552-5260

A Multivariate Genome‐Wide Association Study of the Shared Genetic Architecture Among Modifiable Risk Factors for Dementia: 590 Loci Associated with Brain Health

Isabelle F Foote, Sara Bandres‐Ciga, Jonggeol Jeff Kim, Chelsea X Alvarado, Hampton L Leonard, Phazha LK Bothongo, Andrew D Grotzinger, Benjamin M Jacobs, Sheena Waters, Ruth Dobson, Alastair J Noyce, Kamaldeep S Bhui, Ania Korszun, Charles R Marshall
  • Psychiatry and Mental health
  • Cellular and Molecular Neuroscience
  • Geriatrics and Gerontology
  • Neurology (clinical)
  • Developmental Neuroscience
  • Health Policy
  • Epidemiology



Targeting modifiable risk factors may prevent or delay dementia. The mechanisms by which some risk factors influence dementia remain unclear and commonality between risk factors is often overlooked. We aimed to characterize the shared genetic architecture of modifiable risk factors for dementia, and use this to explore whether shared pathways influence dementia pathogenesis and brain health.


We performed two multivariate genome‐wide association studies (GWAS) to characterize the shared genetic architecture of 12 modifiable risk factors for dementia using genomic structural equation modeling (GSEM). One GWAS measured genetic associations with 3 latent subgroups of closely correlated risk factors, whereas our second GWAS measured genetic associations shared across all 12 risk factors (Figure 1). We tested whether the associated variants were enriched in neuronal/glial subtypes and epigenetic marks found in different brain regions using stratified‐GSEM and identified potentially causal genes for each of the latent factors using gene mapping techniques and summary‐data‐based Mendelian randomization. Pathway analysis was conducted to examine whether the prioritized genes were implicated in dementia‐related biological pathways.


We identified 305 genome‐wide significant loci for the 3‐factor model of modifiable dementia risk and 285 genome‐wide significant loci for the Common Factor. Thirty‐four of these loci were novel (not previously associated with any phenotype by GWAS). Post‐GWAS analyses uncovered multiple links between the shared risk factor loci and dementia pathogenesis. Stratified‐GSEM showed that shared risk factor variants are significantly enriched in regions of the brain that degenerate in dementia, including the hippocampus and prefrontal cortex. Gene mapping identified potentially shared causal genes between risk factor loci and Alzheimer’s disease (AD) loci. Pathway analysis indicated that our prioritized genes were enriched in AD pathways, protein aggregation and neuronal cell signaling. Mendelian randomization provided evidence for a causal role of the shared risk factor variants in AD pathogenesis.


We highlight multiple genomic regions where genetic overlap between modifiable risk factors seems to directly influence dementia‐related pathways in regions of the brain that are known to degenerate in dementia. Our findings provide novel insights into the potential biological mechanisms by which modifiable risk factors may increase risk for dementia.

More from our Archive