Alzheimer’s Gene Doesn’t Behave the Same Way in Everyone

On a PET scan, amyloid shows up as a wash of colour across the cortex, brighter where the sticky protein has gathered into plaques. For decades the reading of those scans has rested on a tidy assumption: carry the APOE ε4 gene variant, the strongest common genetic risk factor for late-onset Alzheimer’s, and your brain is far more likely to light up. More amyloid, more risk. The logic felt close to mechanical. A new analysis of more than 17,000 older adults suggests the machinery is not so uniform after all.

Researchers at the USC Mark and Mary Stevens Neuroimaging and Informatics Institute pulled together brain imaging and clinical records from five large aging and Alzheimer’s studies. What they found, published this month in Alzheimer’s & Dementia, complicates one of the field’s load-bearing relationships, and it does so along lines of ethnicity that the research has too often skated past.

The trick that made the work possible is a number called the Centiloid. Different studies use different radioactive tracers, different scanners, different timing windows, and for years that meant amyloid measurements simply could not be stacked side by side. The Centiloid scale fixes a common ruler over all of it, anchored at zero for a clean amyloid-negative brain and at 100 for the typical scan of someone with Alzheimer’s dementia. Convert everything to Centiloids and, suddenly, seventeen thousand scans from across the United States and Canada can be read together.

That pooling is not a minor convenience. Hispanic participants made up only 8.4 per cent of the combined dataset, just 1,427 people, which is roughly the problem in miniature: no single study had enough of them to say anything firm.

“This is exactly the kind of work GAAIN was built to enable,” says Arthur Toga, director of the Stevens institute and the study’s senior author, referring to the Alzheimer’s data-sharing platform that let his team locate and link the compatible datasets. The point of the scale, and the network behind it, is reach. Patterns invisible inside one cohort start to surface once the cohorts are combined.

A weaker link than the textbook says

And the pattern that surfaced was this. Across both Hispanic and non-Hispanic white participants, more amyloid still tracked with worse cognition and with carrying APOE ε4, exactly as expected. But Hispanic participants, at every level of diagnosis, carried less amyloid than their non-Hispanic white counterparts with the same profile.

The gene’s grip looked looser, too. Non-Hispanic white carriers of APOE ε4 were more than four times as likely to show amyloid pathology in their brains; for Hispanic carriers the figure was about two and a half times. Among Hispanic participants diagnosed with mild cognitive impairment, the variant was not significantly tied to higher amyloid at all. The risk factor everyone treats as a near-constant turned out to be doing rather less heavy lifting in one group.

“APOE ε4 is a major Alzheimer’s disease genetic risk factor, but our results suggest its relationship to amyloid buildup may be more nuanced in Hispanic populations,” says Cally Xiao, the study’s lead author. Nuanced is doing a lot of work in that sentence, and deliberately so; the data describe a difference, not yet a cause.

Here is where the finding could be badly misread, and the authors are plainly anxious that it might be. Less amyloid sounds like less danger. It is not. Hispanic populations actually face a higher overall burden of dementia, which is precisely what makes lower amyloid so puzzling rather than reassuring. If the plaques aren’t there in the same measure, something else is driving the decline. Vascular problems, perhaps. Other proteins. Social and economic forces that imaging cannot see. The team checked one obvious candidate by building a composite vascular score from conditions like diabetes, hypertension and stroke, all more common among the Hispanic participants. Adjusting for it softened the ethnic gap but did not erase it. So vascular health is part of the story. Not the whole of it.

“These findings do not mean that Hispanic adults are at lower risk for dementia,” Xiao says. “In fact, Hispanic populations face a higher overall burden of dementia. Instead, our results suggest that cognitive impairment in Hispanic older adults may not always be driven by amyloid in the same way, and that other biological, vascular, or social factors may also be important.”

The caveats are real and the authors list them without flinching. The Hispanic sample, though large by the standards of this kind of work, was still a fraction of the whole, and “Hispanic” itself flattens an enormous diaspora with different ancestries and histories. The analysis was a snapshot rather than a film; nobody followed these brains across years of change. And the participants came from studies built for other purposes, with their own quirks of recruitment.

Why the timing bites

What sharpens all of this is the moment it arrives in. A new class of anti-amyloid drugs is reshaping Alzheimer’s treatment, and the entire premise of those therapies is that clearing plaques will help. If amyloid is not the same engine of decline in every population, then a treatment aimed squarely at amyloid may not deliver the same benefit to everyone who takes it, an uncomfortable possibility for a field whose trials have long been dominated by non-Hispanic white volunteers.

“Alzheimer’s disease is complex, and the path to cognitive decline may not look identical for every population,” Toga says. The next step is the obvious hard one: more Hispanic participants, scanned and then followed over time, so that a difference glimpsed in a single snapshot can be watched to see where it actually leads.

Source: Xiao C, et al. Association of cognitive impairment and APOE ε4 with Centiloids in Hispanic and non-Hispanic White cohorts. Alzheimer’s & Dementia (2026). https://doi.org/10.1002/alz.71586

Frequently Asked Questions

If Hispanic adults have less amyloid in their brains, why do they face more dementia overall?

That apparent contradiction is the whole point of the study. Lower amyloid does not mean lower risk; it suggests the decline may be driven by something other than plaques, such as vascular damage, other proteins, or social and economic factors that imaging cannot capture. Untangling exactly what fills that gap is the open question the researchers want answered next.

Does this mean the APOE ε4 gene doesn’t matter for Hispanic people?

No, it still raises risk, just less steeply than in non-Hispanic white people. White carriers were more than four times as likely to show amyloid pathology, while Hispanic carriers were about two and a half times as likely, and in one group the link disappeared entirely. The gene’s influence appears real but weaker, which is itself a clue worth chasing.

Could the new anti-amyloid Alzheimer’s drugs work less well for some patients?

It is a genuine possibility the findings raise. Those therapies are built on the premise that clearing amyloid will slow decline, so if amyloid is not the main driver in a given population, the benefit could differ. The study does not test the drugs directly, but it flags a question the trials, long dominated by white volunteers, will need to confront.

How can researchers compare brain scans from dozens of different studies?

They use a standardized measure called the Centiloid scale, which converts readings from different tracers, scanners and methods onto one common ruler anchored from zero to 100. That harmonization is what let the team pool more than 17,000 scans that would otherwise have been impossible to compare. It is a quiet bit of plumbing that is reshaping what large-scale brain research can ask.