Am J Geriatr Psychiatry. 2025 Oct 22:S1064-7481(25)00500-7. doi: 10.1016/j.jagp.2025.10.005. Online ahead of print.
ABSTRACT
OBJECTIVES: Attention-deficit/hyperactivity disorder (ADHD) has been associated with increased risk for dementia, including Alzheimer’s disease (AD). Polygenic risk scores for ADHD (ADHD-PRS) reflect the genetic liability for the disorder and have been linked to cognitive decline in selected clinical samples. This study examined associations between ADHD-PRS and cognitive function in older adults from a population-based cohort and tested whether these associations were moderated by plasma biomarkers of AD and related disorders (ADRD).
METHODS: We analyzed data from 1,468 dementia-free participants from the Monongahela-Youghiogheny Healthy Aging Team (MYHAT) study, with cognitive assessments across five domains. ADHD-PRS was calculated, and a subset underwent blood draw for the analysis of ADRD biomarkers. We examined cross-sectional and longitudinal associations between ADHD-PRS and cognition using linear regressions and mixed-effects models adjusting for age, sex, and ancestry.
RESULTS: Higher ADHD-PRS was significantly associated with lower visuospatial function in cross-sectional analyses after correction for multiple comparisons. Associations with attention and language were moderated by education, and higher ADHD-PRS was associated with lower performance only among individuals with ≤ high school education. No associations were found between ADHD-PRS and plasma biomarkers, nor did biomarkers modify the relationship between ADHD-PRS and cognition. In longitudinal analyses, no association between ADHD-PRS and cognitive trajectories remained significant after correction for multiple comparisons.
CONCLUSIONS: Overall, ADHD genetic liability appears relevant to cognitive performance in late life, though longitudinal trajectories remain unaffected. In older age, ADHD-PRS seems to negatively impact visuospatial tests, characterized by complex and integrative demands. Educational attainment moderated the associations with attention and language, suggesting a potential buffering effect.
PMID:41241612 | DOI:10.1016/j.jagp.2025.10.005
