Developed by researchers at the Washington University School of Medicine in Saint-Louis, this experimental test analyzes fifteen proteins present in the blood in order to identify the biological signatures of four major neurodegenerative diseases. Although promising, it remains at the research stage and will still need to be validated in larger populations before entering clinical practice.
Behind the symptoms, a diagnosis often more complex than it seems
For patients and their loved ones, the first signs of dementia are often destabilizing. An unusual oversight. Difficulty finding words. Behavioral changes. Little by little, worry sets in and with it an essential question: what disease is it really about? Because contrary to popular belief, not all dementias are the same.
Alzheimer’s disease, Parkinson’s disease associated with cognitive decline, frontotemporal dementia or dementia with Lewy bodies can cause overlapping symptoms. Added to this is another difficulty: several pathologies can coexist in the same person.
Today, establishing an accurate diagnosis often relies on a combination of specialized tests, sometimes cumbersome or expensive, such as positron emission tomography (PET scan) or lumbar punctures.
For Carlos Cruchaga, director of the NeuroGenomics and Informatics Center at the Washington University School of Medicine, this situation considerably limits the understanding of what is really happening in patients’ brains.
“Right now, many patients are given a single diagnostic label, such as Alzheimer’s or Parkinson’s, but in reality their brains often have a mixture of lesions. Current tools simply weren’t designed to detect this.”he explains in a press release from the Washington University School of Medicine.
Faced with this complexity, researchers wanted to develop a tool capable of going beyond traditional categories.
“Our goal was to create a test that doesn’t just say “yes” or “no” to a disease, but gives an indication of all the major neurodegenerative diseases present in a person.”continues Carlos Cruchaga.
This ambition is at the heart of the GPND AI-NULISA project, presented in the journal Alzheimer’s & Dementia.
Fifteen proteins and artificial intelligence to decipher the brain
The idea may seem surprising: understanding what is happening in the brain from a simple blood test. However, neurodegenerative diseases leave biological traces in the body. Certain proteins circulating in the blood reflect processes of neuronal degeneration, inflammation or even damage to connections between neurons.
The test developed by the American team analyzes 15 of these biomarkers simultaneously. Among them is the phosphorylated protein tau 217 (p-tau217), today considered one of the most effective blood indicators of Alzheimer’s disease. But the researchers chose to go further by also integrating proteins associated with nerve damage, synapses and inflammatory phenomena.
This data is then interpreted by an artificial intelligence model designed to estimate the probability of five different situations: Alzheimer’s disease, Parkinson’s disease, frontotemporal dementia, dementia with Lewy bodies or brain aging without major neurodegenerative disease.
And the performances obtained are remarkable! The model was developed using data from 3,233 participants followed in specialized American centers. It was then evaluated in several independent cohorts, including a group of 225 people whose brains could be studied after their death in order to compare the algorithm’s predictions to biological reality. Result: an overall diagnostic accuracy of 92.3%.
The researchers also observed that people identified by artificial intelligence as having a high probability of Alzheimer’s disease actually had more amyloid plaques and tau lesions at autopsy. They also had more impaired cognitive performance.
Even more interesting, the test was able to detect biological signatures consistent with Alzheimer’s disease in some patients who were initially diagnosed with Parkinson’s and who subsequently developed dementia.
These results suggest that certain brain transformations could be identified well before they become clinically evident.
A promise for tomorrow, but not yet a routine exam
As encouraging as they are, these results do not mean that this test will be available in medical offices in the coming months. Before possible clinical use, the tool will need to be evaluated in larger, more diverse populations and followed over an extended period. The objective will be to verify its ability not only to identify diseases, but also to predict their evolution and guide therapeutic decisions.
For the moment, it is therefore a research tool. But in a context where the population is aging and where the number of people living with dementia continues to increase throughout the world, having a minimally invasive blood test could make it possible to refer patients more quickly to the right specialists, to select the additional tests that are really necessary or even to facilitate access to targeted clinical trials.
This approach could also support the development of more personalized medicine for neurodegenerative diseases, based on the biological mechanisms specific to each patient rather than on a simple diagnostic category. For families faced with the onset of cognitive disorders, the issue goes beyond technological performance: this would make it possible to understand what is happening earlier, reduce uncertainty and allow appropriate care at a time when it can still make a difference.
If the promises of this test are confirmed in the years to come, it could mark an important milestone in the history of dementia diagnosis. Not by replacing the doctor, but by giving him a new tool to see more clearly what, until now, often remained difficult to distinguish.