Altered protein shapes in the blood can reveal early stages of Alzheimer’s disease

Researchers have developed a new blood test that detects Alzheimer’s disease by analyzing the physical shapes of proteins rather than just counting their quantities. This structural approach accurately identifies the disease stages and offers fresh insights into how genetic risks and behavioral symptoms differ between men and women. The findings were recently published in the journal Nature Aging.

Proper cellular function relies on a strict quality control system that folds proteins into precise three-dimensional shapes. When this system fails as people age, misfolded proteins can build up and disrupt normal biological processes. In conditions like Alzheimer’s disease, defective proteins accumulate in the brain years before memory loss or other cognitive issues become visible.

Historically, most diagnostic blood tests have focused on measuring the total concentration of specific disease-linked proteins. The research team suspected that examining the physical structures of these proteins might reveal far more about the disease mechanism than expression levels alone. They reasoned that tracking how protein shapes change in the bloodstream could provide early warning signs of cognitive decline.

The research was led by Ahrum Son and supervised by John R. Yates III at The Scripps Research Institute in California. They collaborated with colleagues from the University of Ulsan, Chungnam National University, the University of Kansas Medical Center, and the University of California, San Diego. The team set out to map the structural modifications of proteins across different stages of memory impairment.

The investigators analyzed blood plasma samples from 520 volunteers. These participants included healthy individuals, people with mild cognitive impairment, and patients diagnosed with Alzheimer’s disease. According to the research announcement, “The individuals were volunteer research participants at the NIA-funded Alzheimer’s Disease Research Centers in Kansas and California, where they were seen for annual visits.”

To examine the protein shapes, the team utilized a chemical tagging technique combined with mass spectrometry. Mass spectrometry is an analytical tool that identifies molecules based on their mass and charge, allowing scientists to see the exact composition of a sample. This combined approach allowed the researchers to profile the proteins circulating in the blood of the participants.

This chemical tagging method measures which parts of a protein are exposed to the surrounding fluid and which parts are buried inside the folded structure. As proteins misfold, their normally hidden sections can become exposed, or their usually exposed parts can become trapped inside. By quantifying these structural shifts across the entire blood sample, the researchers created a broad profile of shape changes associated with cognitive decline.

The team first looked at a protein called Apolipoprotein E, which acts as a carrier for cholesterol in the blood. Humans typically inherit one of three genetic variations for this protein. One specific variant is known to heavily increase the risk of developing Alzheimer’s disease.

The researchers mapped how these genetic differences physically altered the shapes of other interacting proteins in the blood. They found that the high-risk genetic variant led to distinct structural configurations in the surrounding protein network. This suggests that the genetic risk might manifest through the physical warping of other essential molecules.

Next, the investigators examined the relationship between protein shapes and neuropsychiatric symptoms, such as anxiety, apathy, and hallucinations. They noted that nearly all patients with Alzheimer’s experience these behavioral changes, but men and women often display different symptom patterns. The researchers explored whether physical protein changes could help explain these observed behavioral differences.

The structural protein profiles revealed distinct differences between the sexes in how physical protein changes correlated with the severity of mood and behavior symptoms. Women in the advanced disease group showed higher scores for cognitive impairment and mood disorders than men. The protein shape changes accurately mirrored these sex-based differences in symptom severity.

To translate these broad biological observations into a practical screening tool, the team tested eighteen different machine learning algorithms. Machine learning involves training a computer program to recognize patterns within large datasets. They sought an automated way to predict a patient’s cognitive status based purely on their blood protein structures.

A deep learning model performed the best among all the computational methods tested. Deep learning uses layered artificial neural networks to process data in ways that mimic the human brain. This algorithm ultimately identified a specific pattern of structural changes in three particular proteins known as C1QA, CLUS, and ApoB.

These three proteins are involved in immune responses, waste clearance, and lipid transport. This three-marker panel successfully distinguished between healthy aging, mild cognitive impairment, and full Alzheimer’s disease with an accuracy of 83.44 percent. The model achieved this without needing any prior information about the patients’ clinical diagnoses.

When looking at just two groups at a time, the model separated healthy individuals from those with mild cognitive impairment with high precision. It was equally accurate at distinguishing mild cognitive impairment from advanced Alzheimer’s disease. The structural changes also correlated with physical changes in the brain, such as the enlargement of brain fluid cavities.

The researchers compared their structural model against a traditional model that only measured protein quantities. The structural model dramatically outperformed the quantity-based approach in categorizing the patients. This suggests that structural data holds more diagnostic power than mere protein abundance.

The researchers also tested the model on 50 participants who were tracked over several months. The tool correctly identified their advancing disease status 86 percent of the time. This confirmed that the protein shape signatures change dynamically as the disease worsens.

While the diagnostic panel shows immense promise, the study does have some limitations. During the blood preparation process, the researchers had to remove highly abundant proteins to detect the rarer ones. This necessary filtering step might have accidentally removed some disease-linked proteins that were physically attached to the abundant ones.

This removal means some helpful biological markers might have been lost before the analysis even began. Additionally, the longitudinal portion of the study tracked a relatively small number of patients for less than a year. The changes observed over this short follow-up period were not statistically significant regarding the prediction of long-term disease trajectories.

The research team noted that tracking larger groups of patients over several years will be required to fully test the clinical utility of this tool. Despite these constraints, the study provides a robust foundation for a new class of diagnostic tests. By focusing on the physical shapes of molecules, clinicians might soon have a more detailed window into the early stages of brain deterioration.

The findings open up entirely new avenues for monitoring how the disease responds to experimental treatments. As researchers work to refine the test, it could become a standard part of routine medical checkups. This would allow doctors to intervene much earlier in the disease process.

The research team shared their excitement about the implications of the new diagnostic test. “With this work, we established a potential new biomarker panel that reveals structural disruptions in proteins linked to Alzheimer’s disease that are invisible to traditional approaches,” said Yates, a professor of Integrative Structural and Computational Biology at The Scripps Research Institute. “This approach accurately distinguishes stages of the disease, meaning that it could help enable earlier diagnosis.”

This research was supported by the National Institutes of Health. The study, “Structural signature of plasma proteins classifies the status of Alzheimer’s disease,” was authored by Ahrum Son, Hyunsoo Kim, Jolene K. Diedrich, Casimir Bamberger, Heather M. Wilkins, Jeffrey M. Burns, Jill K. Morris, Robert A. Rissman, Russell H. Swerdlow & John R. Yates III.