A new class of Alzheimer’s biomarkers: Why protein shape may beat protein levels

Workflow of CPP using the blood samples. Credit: Nature Aging (2026). DOI: 10.1038/s43587-026-01078-2

Researchers have identified a new type of blood-based biomarker test for Alzheimer’s disease that measures structural changes in proteins, providing more information on the underlying biology of the disease than standard blood tests. The findings, published in Nature Aging, also provide new insights into how Alzheimer’s disease biology may differ between males and females.

“This work introduces a fundamentally new, blood-based approach to detecting and staging Alzheimer’s disease,” said Dr. Richard Hodes, director of NIH’s National Institute on Aging (NIA). “By revealing protein structural changes associated with genetic risk, symptom severity, and sex differences—features not captured by existing biomarkers—this research could enable earlier diagnosis and more effective clinical trials.”

Looking beyond traditional protein measures

Most Alzheimer’s blood tests measure how much of an Alzheimer’s-linked protein is present. However, it is known that in Alzheimer’s disease, breakdown of the regulation of cell function causes protein misfolding. The researchers wanted to know whether these structural changes could be identified in blood tests to detect the disease.

They hypothesized that comprehensive investigation of structural changes in Alzheimer’s-associated proteins could reveal more about the mechanisms underlying disease risk factors and symptoms than current blood tests and could potentially identify additional blood-based biomarkers for the disease.

In addition, almost all individuals with Alzheimer’s develop neuropsychiatric symptoms, but research suggests differences between males and females in the frequency and severity of certain symptoms. The authors wondered whether structural changes in proteins could help researchers better understand the biological processes underlying these sex differences.

To address these questions, the researchers analyzed blood plasma samples from 520 individuals, including people with diagnosed Alzheimer’s, people with mild cognitive impairment, and healthy controls. The individuals were volunteer research participants at the Alzheimer’s Disease Research Centers in Kansas and California, where they were seen for annual visits.

Machine learning reveals structural protein signatures

Using mass spectrometry and machine learning, the researchers were able to characterize changes in protein structure associated with genetic risk for Alzheimer’s—specifically in variants of the ApoE gene. They also connected disease-related changes to the severity of neuropsychiatric symptoms in males and females, observing distinct structural patterns by sex.

The research team then used machine learning to develop a diagnostic panel of three proteins—C1QA, CLUS, and ApoB—representing Alzheimer’s-associated structural changes. They found that the panel could accurately distinguish between Alzheimer’s, mild cognitive impairment, and healthy controls, and it could distinguish disease stages and track progression of disease over time.

Potential impact on diagnosis and trials

“With this work, we established a potential new biomarker panel that reveals structural disruptions in proteins linked to Alzheimer’s disease that are invisible to traditional approaches,” said Dr. John Yates, lead author of the study and professor of Integrative Structural and Computational Biology at The Scripps Research Institute in La Jolla, California.

“This approach accurately distinguishes stages of the disease, meaning that it could help enable earlier diagnosis.”