Common Genes Carry Nearly Triple the Alzheimer's Risk We Thought

The sample was 91% European by effective size, with 5% African and 3% East Asian ancestry. In short: the findings skew heavily toward one population.

What the genes point to

Of the 127 regions, 118 emerged from multi-ancestry analysis and nine from European-only scans. Comparing the results with single-cell RNA data from donated brains showed risk concentrating in genes overexpressed in microglia and underexpressed in neurons. After statistical adjustment, the microglial signal held firm while individual neuronal subtype signals weakened, suggesting broad neuronal dampening rather than damage to specific cell types. The authors cannot yet say whether that dampening triggers disease or follows from it.

Pathway analyses highlighted immune activation, fat binding, cell adhesion, and amyloid clearance, plus five processes new to Alzheimer's genetic scans — including oxidative-stress regulation through glutathione metabolism.

Bigger heritability, sharper prediction

Using a model that accommodates APOE's outsized effect, the team estimated common variants explain 16% of population variation in susceptibility among Europeans, assuming 5% lifetime prevalence — nearly triple the older 6% figure, which simulations show understates heritability when one gene dominates.

A polygenic risk score — a single number summarizing inherited risk across millions of variants — explained nearly 13% of susceptibility variation in held-out European cohorts, beating APOE status alone. In UK Biobank adults 65 and older, prevalence was seven times higher in the top 10% of scores than the bottom 10%.

Uffelmann put it in absolute terms: "In people with the highest genetic risk (top 1%), the fraction of Alzheimer's cases increased to almost 1 in 4. Clearly, genetics plays an important role in affecting risk, but it does not determine diagnosis."

The high-risk ε4 variant of APOE showed a roughly 13% stronger association in women than men; otherwise inherited risk overlapped closely between sexes.

Drug targets and shared pathways

Cross-referencing new regions with regulatory databases flagged 17 genes that interact with approved non-Alzheimer's drugs; compounds targeting three showed promise in animal studies, while two others had already failed in mice or humans. The model newly prioritized QPCT, an enzyme that shapes amyloid-beta into the form targeted by the approved antibody donanemab. A new region spanning UBE2V1 and SPATA2 shares a likely causal variant with ALS, hinting at shared disease mechanisms.

Caveats practitioners should weigh

Several co-authors hold employment or stock in drugmakers including Roche, Eli Lilly, and Regeneron; co-author Kaj Blennow joined Eli Lilly in September 2025. Roughly 40% of cases — 62% to 82% in non-European groups — relied on self-reported parental dementia, which may blend Alzheimer's with other dementias. Polygenic scores performed far worse in African and East Asian cohorts, and a genetic match does not prove a drug target will slow decline.

As Uffelmann stressed: "It's only part of the story, and lifestyle and environmental factors also matter.