A blood test told three dementias apart, not just detected one

The recent run of Alzheimer’s blood tests answers a yes-or-no question. Someone sitting in a memory clinic usually needs a different one answered: not whether something is wrong, but which thing.

A study in Nature Aging went after that harder question. The authors note that blood-based biomarkers are emerging as scalable tools for the diagnosis and monitoring of neurodegenerative diseases, but markers enabling differential diagnosis across major dementias remain limited.

Why the distinction is the clinical problem

Dementia is not one disease, and the differences are not academic.

Alzheimer’s disease, dementia with Lewy bodies and frontotemporal dementia progress differently, respond to different drugs, and demand different care planning. Lewy body disease carries a specific hazard: severe reactions to certain antipsychotic medications, which makes misidentifying it actively dangerous.

Diagnoses made from symptoms alone are wrong often enough to matter, and the definitive answer has historically arrived at autopsy. A blood test that separated the three during life would change what a first appointment can accomplish.

What was measured

The scale here is what makes it credible. Large-scale plasma proteomics was used to identify disease-associated signatures across Alzheimer’s disease, dementia with Lewy bodies and frontotemporal dementia.

The researchers analyzed 1,318 plasma samples from well-characterized international cohorts and identified more than 200 dysregulated proteins across disease groups.

Two hundred proteins is a wide net, and casting a wide net invites false positives. What follows is the part that addresses it.

The signatures for each disease

Each of the three conditions came with a distinguishable pattern rather than a single marker.

For Alzheimer’s, glial fibrillary acidic protein showed the strongest increase along the Alzheimer’s disease continuum. That protein comes from astrocytes, the brain’s support cells, and rises as they react to accumulating damage.

For Lewy body disease, integrin alpha-V and integrin alpha-M were consistently reduced in Lewy body disorders, including autopsy-confirmed cases. The autopsy detail is the most important four words in the paper. A marker validated only against diagnoses made in life can learn the mistakes in those diagnoses; one that holds in autopsy-confirmed cases is tracking the disease as pathology defines it.

For frontotemporal dementia, elevated neurofilament light chain and lower glial fibrillary acidic protein were associated with frontotemporal dementia. Neurofilament light chain is structural material from inside nerve fibers, released when axons break down.

Notice the shape of that third signature. It is not a marker being high; it is one marker high while another is low. That combination is why a panel can separate diseases where any single number cannot.

The step that turns a finding into a tool

Two hundred proteins is a research result. A usable test needs to be smaller.

The researchers translated these findings into a 21-protein quantitative multiplex panel and validated it in an independent multicenter cohort of 805 people.

Independent validation is the check most biomarker papers skip, and skipping it is why so many promising panels never survive contact with a second dataset. Doing it here, in a separate multicenter group, is the difference between a signal and a candidate test.

What this is not

There is nothing to ask for at an appointment. This is a research panel run in specialist laboratories on samples from established cohorts.

The larger caveat is who was in those cohorts. The samples came from well-characterized international cohorts - people already investigated thoroughly enough to carry a confident diagnosis. Actual memory clinics see earlier presentations, vaguer symptoms, and a great many older people with more than one pathology at once. A panel that cleanly separates textbook cases may blur on exactly the mixed ones where a clinician most wants help.

The authors’ framing is appropriately narrow: these findings support plasma proteomics as an approach for biomarker-based differential diagnosis and disease staging across major neurodegenerative dementias.

An approach, then, rather than a product. What makes it worth reporting is the ambition of the question. Detecting Alzheimer’s from blood was the achievement of the last few years. Telling it apart from its neighbors is the one that would change a Tuesday morning in a memory clinic.