Researchers at the Karlsruhe Institute of Technology (KIT) have developed a new method to detect neurodegenerative diseases. The method is effective and simple and can detect the misfolding of proteins underlying the disease at an early stage of the disease. The prediction accuracy is more than 99 percent, announced the KIT. The associated study was published in April in the specialist journal “Advanced Materials”.
Neurodegenerative diseases such as Alzheimer’s or Parkinson’s are caused by misfolding of proteins or peptides, i.e. by changes in their spatial structure. The cause is the smallest deviation in the chemical composition of the biomolecules. According to a study on the drying structure of protein and peptide solutions, these misfoldings can be recognized.
To do this, the researchers led by Professor Jörg Lahann analyzed microscopic images of such solutions and evaluated them with neural deep learning networks. The artificial intelligence (AI) was able to derive the underlying biochemical structure from the stains left by the drying droplets of the peptide solutions on a solid surface. The spot patterns are difficult to distinguish with the naked eye, but are like “fingerprints” for the AI, reflecting the structural and spatial identity of a peptide.
In this way, the researchers classified eight genetic mutations that lead to misfolding in proteins that are known to be involved in neurodegenerative diseases and can be effectively detected using the method. The technology allow to identify Alzheimer’s variants within minutes. Because no time-consuming sample preparation is required, the method allows simple and patient-oriented diagnostics.
In principle, any doctor might use the method and even read the protein patterns with a smartphone, Lahann told “Research & Teaching”. A clinical application is not foreseeable at the moment, further studies are necessary. According to Lahann, the peptides used in the study correspond to different variants of a peptide that exist in patients and represent different genetic mutations. It still has to be researched whether the method also works reliably with real patient samples.