The pattern reveals it: In the future, the pattern of a drying sample droplet could reveal whether someone has Alzheimer’s in the early stages. Researchers have discovered that if the sample contains misfolded amyloid beta proteins, this changes the drying pattern of the solution in a characteristic way. In the first tests, an algorithm trained on this recognized this “fingerprint” with 99 percent accuracy – and this without the need for complex processing and analysis methods.
Alzheimer’s is the most common neurodegenerative disease in old age. However, by the time a diagnosis can be made, the breakdown of the brain cells is often far advanced. This is because there are currently no objective, easy-to-test biomarkers for early detection. It is true that early signs of dementia can be identified from the brain waves intestinal flora or blood tests recognize. However, these methods are still in the testing stage and often require complex procedures and equipment.
Dry spots as indicators?
But there may also be an easier way: Azam Jeihanipour and Jörg Lahann from the Karlsruhe Institute of Technology (KIT) have developed a method that could easily indicate the presence of pathologically altered amyloid beta proteins in blood or cerebrospinal fluid samples. Such misfolded beta-amyloids are considered indicators of Alzheimer’s. They cannot be broken down properly in the brain, so they form clumps that damage and kill neurons.
The so-called form the basis of the new method “coffee ring” stains – the stains left by a dried solution on a surface. The structure of such dry spots is strongly influenced by the chemical properties of the ingredients – and thus also by the three-dimensional folding and structure of the proteins. “Published speckle patterns of peptide and protein solutions range from homogeneous films to branched and latticed patterns to more complex arrangements,” explain Jeihanipour and Lahann.
Amyloid droplets under test
In order to find out whether healthy and diseased amyloid beta variants can also be distinguished based on this drying pattern, the researchers dissolved various amyloid variants in a bicarbonate buffer solution and applied two microliters of each as drops to a previously specially coated glass surface. The drops were then dried for 40 minutes under controlled conditions.
The result is dry spots around two millimeters in size, which show a characteristic pattern under the polarizing microscope. Salt crystals from the buffer solution are predominantly deposited in the center of the spots, while peptides and salt crystals form a relatively homogeneous border on the outer edge. In the middle zone, however, branching structures can be seen that extend from the inside outwards and appeared to be specific for each peptide variant.
“However, recognizing the differences between these patterns with the naked eye is quite a challenge because they look very similar,” the researchers explain. They have therefore enlisted the help of an adaptive algorithm. This deep learning system was first trained with around 400 spot pattern recordings per amyloid variant. Then the algorithm should automatically classify 720 new images of the eight peptide configurations.
99 percent accuracy
And indeed: The AI system was able to recognize and correctly assign the structure and folding of the beta-amyloid chains solely from the dry spots. “The speckle patterns were not only characteristic and reproducible, but also led to a classification with a prediction accuracy of over 99 percent,” reports Lahann. “Since the structures are very similar and difficult to distinguish with the naked eye, it was quite surprising that the neural networks were so effective.”
According to the researchers, this demonstrates that the various amyloid variants relevant to Alzheimer’s can also be identified using this comparatively simple and fast method. “The speckle patterns of amyloid-beta peptides represent fingerprints that reflect the structural and spatial identity of the peptide,” says Lahann.
This method therefore has great potential to be used as a quick and simple test for the early detection of Alzheimer’s and Parkinson’s. (Advanced Materials, 2022; doi: 10.1002/adma.202110404)
Source: Karlsruhe Institute of Technology