A study describing a new class of Alzheimer's disease biomarkers, based on measuring structural changes in blood proteins rather than how much of a protein is present, was announced by the National Institutes of Health and published as a peer-reviewed paper. Researchers profiled plasma protein structures from 520 participants, including people with Alzheimer's disease, people with mild cognitive impairment (MCI), and healthy controls, drawing on blood samples collected through Alzheimer's disease research centers at the University of California San Diego and the University of Southern California. Using mass spectrometry combined with machine learning, the team characterized structural changes across the plasma proteome and identified a diagnostic panel built from peptides of three proteins: C1QA, CLUS and ApoB. In three-way classification distinguishing healthy participants, those with MCI, and those with Alzheimer's disease, the panel achieved 83.44 percent accuracy, according to the published findings.
Most blood-based Alzheimer's tests developed to date, including those already in clinical use, measure the concentration of proteins such as amyloid or tau, essentially asking how much of a substance is present. This study instead measured protein conformation, the physical shape a protein folds into, on the premise that Alzheimer's disease involves a breakdown in normal protein folding (proteostasis) that can alter a protein's structure even before or independent of changes in its overall level. The researchers describe this structural proteomics approach as capturing information that concentration-based tests miss, and note it also incorporated data on ApoE genetic variation and neuropsychiatric symptoms in refining the disease-associated signatures.
The finding is observational and cross-sectional in the sense that it classified participants at a single point based on existing diagnostic categories (established through cerebrospinal fluid analysis for amyloid and tau) rather than following people over time to see whether the structural markers predict who will develop Alzheimer's disease. The 83.44 percent accuracy figure describes performance within this study's own cohort; the authors themselves frame the panel as holding promise for early detection and stage classification, language that signals a research tool at an early stage rather than a diagnostic ready for clinic use. Independent replication in separate cohorts, and testing of whether the structural signature appears before symptoms, has not yet been reported.