17,000 Brain Scans Reveal Ethnic Differences in Amyloid

Think of Alzheimer’s research as a vast jigsaw puzzle. For years, many pieces fit neatly for some populations but left puzzling gaps for others. A new, large-scale imaging analysis now pulls together scans and clinical data from more than 17,000 older adults and shows those gaps are not random — they point to real biological differences tied to ethnicity and genetics.

Why a harmonized view of amyloid matters

Amyloid plaques are central to most modern explanations of Alzheimer’s disease, and much of current drug development targets amyloid removal. But can we assume the link between plaques, cognitive decline, and genetic risk is uniform across diverse populations? A team at the USC Mark and Mary Stevens Neuroimaging and Informatics Institute set out to test that assumption.

Using the Centiloid scale — a standardized system that puts amyloid PET scans from different tracers and scanners on the same metric — the researchers pooled data from five major studies. The dataset included 17,017 participants in total, of whom 1,427 self-identified as Hispanic. That breadth made it possible to spot patterns invisible inside any single cohort.

The Centiloid scale allows researchers to combine and analyze amyloid PET imaging data from multiple sources that used different tracers, scanners, and acquisition times. 

Key findings that change how we think about risk

Across diagnostic groups, carrying the APOE ε4 variant still correlated with higher amyloid levels, and people with mild cognitive impairment or dementia generally showed more amyloid than cognitively healthy peers. But the strength of those associations varied by ethnicity.

Hispanic participants consistently had lower average amyloid burden than non-Hispanic white participants who shared similar age, education level, cognitive status, and APOE ε4 carrier status. Among APOE ε4 carriers, Hispanic adults with normal cognition or mild cognitive impairment showed lower amyloid than non-Hispanic white carriers.

Perhaps the most striking statistic: non-Hispanic white carriers of APOE ε4 were more than four times as likely to show evidence of amyloid pathology, while Hispanic carriers were roughly two and a half times as likely. That difference suggests APOE ε4’s effect on amyloid accumulation is attenuated in the Hispanic group, at least within this pooled sample.

What could explain the discrepancy?

Interpretation requires caution. Lower amyloid levels in Hispanic participants do not mean lower dementia risk. Epidemiological data indicate Hispanic communities often face a higher overall load of dementia. Instead, the new analysis implies that cognitive decline in some groups may be driven by a broader mix of causes — vascular disease, metabolic factors, or social determinants such as disparities in access to care and education.

Genetic ancestry is another important variable. Hispanic populations are genetically diverse, with varying proportions of Indigenous American, European, and African ancestry. These ancestral differences can influence how genetic risk factors like APOE ε4 manifest biologically. The study authors call for finer-grained ancestry data and more participants from underrepresented groups to untangle those threads.

How the study was put together

This work depended on the Global Alzheimer’s Association Interactive Network, or GAAIN, a data-sharing platform developed at the Stevens Institute. GAAIN enabled researchers to find and harmonize compatible datasets from five major research efforts, including the Alzheimer’s Disease Neuroimaging Initiative and other large imaging studies. By converting amyloid results to Centiloid units, the group could combine scans from different tracers and centers and run analyses at a scale few individual teams can achieve.

Clinical and research implications

The arrival of anti-amyloid therapies makes these findings especially relevant. If amyloid is not the dominant driver of cognitive symptoms in some populations, then treatments that clear plaques may yield different outcomes across ethnic groups. That does not invalidate anti-amyloid approaches, but it does argue for stratified clinical trials and for complementary strategies that address vascular health, metabolic disease, and social determinants.

Longitudinal follow-up will also be essential. Tracking how amyloid levels and cognition change over years — and how other biomarkers emerge in parallel — will clarify whether observed ethnic differences reflect delayed amyloid accumulation, alternative disease pathways, or interactions with non-amyloid drivers of decline.

Expert Insight

"This study shows why diversity in research samples is not optional — it is central to understanding disease biology and to delivering effective care," says Dr. Elena Márquez, a clinical neuroepidemiologist at a major teaching hospital who was not involved in the study. "We need trials that are representative and biomarker strategies that are tuned to population-specific risk profiles. Otherwise we risk offering therapies that work well for some groups and poorly for others."

Dr. Márquez adds that improving vascular risk management and access to longitudinal imaging in underrepresented communities could yield immediate public health benefits while the field continues to refine targeted therapies.

This research warns against assuming a single biological pathway for Alzheimer’s across populations.

Conclusion

The pooled analysis of more than 17,000 brain scans reframes a simple model of Alzheimer’s as a one-size-fits-all amyloid problem. Instead, the findings press researchers and clinicians to consider population-specific biology, ancestry, and social context when interpreting biomarkers and planning interventions. More inclusive, long-term research will be essential to ensure advances in diagnosis and therapy benefit all communities equitably.