Just taking a few samples of blood could help doctors get a better perspective on the extent of Alzheimer’s disease in the brain soon. Researchers from the University of Gothenburg, and other teams, have discovered a set of seven proteins that when used together, can enhance the accuracy of current blood tests at detecting severe cases of the disease. The discovery, published in JAMA Neurology, offer an easy-to-use compromise for determining which stage of the disease a patient is in and advising treatment.
Blood-based assays, as they are presently designed around the biomarker p-tau217, have already made diagnosis of the disease more facile. Although they are effective at an index of whether or not toxic pathology is present, they still do not give information on the degree to which that pathology is present. PET scans are the current gold standards for high-resolution imaging of tau deposits in vivo, but are a prohibitively expensive, time-consuming, and inaccessible technique.
The new work tackles the task of going beyond that one protein, saying. They then brought in machine learning to analyze large-scale proteomics data from two separate, international groups. They used a new immunoassay platform that allows for simultaneous measurement of over 120 inflammation and neuronal markers from one sample, looking for groups that may help improve staging. Seven-protein signature with the inclusion of p-tau217 was identified and when used in a larger panel in people with existing amyloid plaques, the prediction of late-stage tau pathology improved dramatically. For doctoral student Guglielmo Di Molfetta, this panel allows the team to identify when an individual starts to fit into a late profile.
That insight has practical clinical importance. If we can pin down the stage more precisely, it will inform treatment decisions, as new therapies come along and may have different effects at different levels of advancement. It also refines the process of determining eligibility for clinical trials by ensuring that participants’ disease biology aligns with the trial parameters.
Rather, this approach is meant to complement these measures. A multi-protein blood-based test could provide a clinical or research alternative to tau-PET, varying along neither cost nor invasiveness while providing valuable staging data. Since it uses a single blood draw, it would be many orders of magnitude less invasive and more scalable to the population than any current imaging-based test.
This scalability is important as the prevalence of Alzheimer disease continues to increase and health systems seek to expand access to accurate diagnosis. This work represents years of advancement in blood-based biomarkers. Early single protein assays led the way for broader screening tools, and now the selected multi-protein signature starts to reveal a more complete biological profile of the disease. Beyond the traditional tau and amyloid markers on the other well-established pathways, novel inflammatory markers and neuronal proteins seem to reveal (at least part of) the disease evolution. Machine learning helped the research group identify the best combination out of dozens of proteins measured.
