Multiomics and health records identify candidate Alzheimer’s drug targets
Published in Bioinformatics, Brain/Neurology, Drugs development & discovery, Journals.
October 9, 2026 — A study published in Nature Neuroscience on October 8 combines human genetics, experimental models and electronic health data to identify candidate therapeutic targets for Alzheimer’s disease.
The researchers used Mendelian randomization and multiomics datasets to nominate 19 druggable targets in European-ancestry analyses and seven in African-ancestry analyses. One highlighted target was EPHX2, associated with a protective protein-altering variant, p.Arg287Gln.
Experiments using patient-derived neurons and cerebral organoids linked the variant to lower EPHX2 and phosphorylated tau levels. Pharmacological EPHX2 inhibition also improved cognitive performance in a mouse model of Alzheimer’s disease.
An additional analysis of records from 111,680 people with mild cognitive impairment identified twelve medications associated with lower subsequent Alzheimer’s incidence. Those associations do not prove that the medications prevent disease, and the animal findings do not establish efficacy in humans.
The study offers a framework for prioritizing targets through several kinds of evidence. It reports candidate mechanisms and repurposing signals, rather than a validated new treatment for patients.
Source: Hou and colleagues, Nature Neuroscience, October 8, 2026.