This study used polygenic risk scores (PRSs) derived from UK Biobank data to explore genetic signatures across a spectrum of neurodegenerative diseases (NDDs). Analyses were restricted to genes reported to be expressed in major brain cell types—microglia, astrocytes, oligodendrocytes and neurons—with the aim of quantifying pathology-specific genetic contributions that might map to particular cell populations.
The authors emphasise that clinicopathological heterogeneity and coexistence of multiple neuropathologies in older individuals make direct assessment of underlying pathological processes in living persons challenging. Using cell-type–restricted PRSs was intended to provide a genetic proxy for pathology-specific processes.
Analyses revealed substantial overlap among the purported cell-type specific gene sets, with shared content in the range of 21–29%. Importantly, genes at or neighbouring the APOE locus were represented across all cell-type gene sets. The presence of APOE-related genes in every cell-type annotation indicates that currently available cell-type specific annotations capture considerable shared genetic variation across brain cell types rather than strictly unique, cell-restricted signals.
When combined, the cell-type specific PRSs explained a large proportion of the genetic liability for NDDs in which dementia is a prominent clinical manifestation. The joint contribution of these cell-type PRSs was reported to account for approximately 60–93% of genetic risk for those disorders, indicating that a substantial part of the heritable component can be captured by gene sets annotated to major brain cell types.
Alzheimer's disease (AD) showed associations across PRSs derived from all evaluated cell types. Notably, PRSs based on oligodendrocyte and astrocyte gene sets demonstrated strong effects that remained significant independent of APOE status. This pattern supports a multi-cellular genetic architecture for AD where risk-related variation is not restricted to neurons alone and where glial cell–related genetic signals contribute even after accounting for APOE.
The study reported that AD PRS and APOE status independently differentiated Parkinson's disease (PD) cases with dementia from those without dementia. This suggests shared genetic components influencing dementia that are detectable across disorders, as well as disease-specific modifiers that influence the presence of dementia within PD.
Parkinson's disease cases without dementia and amyotrophic lateral sclerosis (ALS) displayed more distinct genetic profiles with limited overlap with other NDDs. These findings indicate that some disorders or clinical subtypes retain relatively specific genetic architectures that are less captured by the shared cell-type gene sets assessed in this work.
Collectively, results reveal both shared and disease-specific genetic contributions across neurodegenerative disorders. The authors point out two persistent challenges: first, imperfect alignment between clinical phenotypes and the underlying neuropathology, and second, extensive overlap of gene expression across major brain cell types. Both factors complicate attempts to attribute genetic risk to discrete cell types and to resolve the biological mechanisms that drive clinical heterogeneity.
The observation that APOE locus genes appear across cell-type annotations underscores the need for caution when interpreting cell-type specific PRSs, since some of the strongest locus-level effects may be distributed across multiple annotated cell populations.
Analyses used UK Biobank resources under application number 109607. The authors stated that all polygenic risk scores—both whole-genome and cell-type specific—and SNP lists for each cell type will be returned to UK Biobank and made available to other researchers with access to the resource. Ethical approvals, participant consent procedures, and reporting standards were affirmed by the authors. No competing interests were declared.
Funding sources declared included the UK Dementia Research Institute and the Medical Research Council, with specific grant identifiers reported in the source report.