Skip to Main Content

Fighting Alzheimer’s disease is a race against time. By the time most patients are diagnosed and treated, their cognitive symptoms have already advanced significantly.

New research, though, pinpoints protein changes in cerebrospinal fluid highly associated with the development of Alzheimer’s disease. A better biological understanding of the brain disorder could foster new treatments and earlier intervention — as early as 20 years before onset of symptoms according to models from the study.


The findings, published Wednesday in Science Translational Medicine, coincide with revised criteria from the Alzheimer’s Association that lean into the need for biology-based diagnosis and staging of the disease. The underlying biology of Alzheimer’s is still a scientific conundrum, but the approval of a drug targeting the disease’s amyloid pathology has sparked a push for blood and other tests to detect proteins associated with disease progression earlier, when treatment can be more successful.

The study tracked upwards of 5,000 proteins to capture a dynamic picture of protein activity in patients with Alzheimer’s. “We take all these different proteins that we measure and we cluster them into these things called modules, that reflect different biological pathways, or processes, or cell types,” said study author Erik Johnson, assistant professor of neurology at Emory University. Those clusters of protein activity can serve as a springboard for both disease prediction and drug treatment experiments, he said.

This work isn’t the first to hunt for patterns within large numbers of Alzheimer’s-associated proteins. But its comprehensive analytical approach was a strength, said Joseph Quinn, a professor of neurology at Oregon Health and Science University. “It looks at the proteomics in an unbiased way without any preconceptions about the disease,” he said. “It does have the potential to give us new insights into some of the biology underlying Alzheimer’s disease.”


The study consisted of 300 participants: 160 with Alzheimer’s and 140 healthy controls. Johnson and his team analyzed the participants’ CSF, the fluid surrounding the brain, for proteins connected to the most common Alzheimer’s pathology: amyloid beta and tau protein aggregations in the brain. Within the data, they found 34 modules highly associated with Alzheimer’s — each associated with different cellular processes.

Two modules were especially likely to be disrupted in patients with Alzheimer’s: One included proteins involved in degrading and reprocessing proteins and organelles through processes like autophagy and ubiquitination, while the other was involved in glycolysis, the processing of sugars.

In the long term, the goal of this kind of protein profiling is to allow for more precise treatment of Alzheimer’s, tailoring prescriptions and care plans to an individual’s unique signature, said Samuel Gandy, a professor of neurology at the Icahn School of Medicine at Mount Sinai who was not involved in the study, in a message to STAT.

Take a patient whose cerebrospinal fluid displayed activation of reprocessing proteins, Johnson explained: “If you had a drug that targets this autophagy/ubiquitination pathway, that drug may work better or not as well depending on how that pathway is changed in that particular person, and not just whether they have AD based on amyloid or tau,” he said.

In one part of the study, Johnson and his colleagues put that concept to the test. They had found that the glycolysis module was most strongly correlated with cognitive decline. Emory researchers were already conducting a clinical trial of atomoxetine as an Alzheimer’s treatment, so the protein researchers piggybacked on the trial to test whether the drug would impact glycolysis dysfunction.

“Lo and behold, we saw that…the module that was most strongly correlated with cognitive function in that 300 person network was significantly reduced by treatment with atomoxetine,” Johnson said.

George Perry, a neuroscientist at the University of Texas San Antonio, said the high overlap in the discovered modules with metabolic processes could also inform lifestyle changes to decrease the risk of Alzheimer’s. “The heart, very similar to the brain, is extremely oxidative in its metabolism, and all of this pinpoints a fairly strong metabolic component: food, exercise, stress reduction,” Perry said. “Those are all things that impinge on metabolism.”

Perry added that the research highlights amyloid as a marker for the disease, noting, however, that the debate around whether amyloid causes Alzheimer’s is still a spirited one.

A second prong of the study focused on proteins affected by apolipoprotein E ε4, the strongest genetic risk factor for Alzheimer’s. Pathways involved in protein modification, cellular signaling, and mitochondrial function were significantly associated with Alzheimer’s patients. Those changes in CSF overlapped substantially with protein expression in blood — which is easier for researchers and clinicians to collect, avoiding an involved lumbar puncture. So in two longitudinal studies, the group looked to find whether these APOE-related proteins were affected in blood the same way they were in CSF.

Both studies collected blood from patients, and then followed up with them over an average of 13 years and 21 years to see who developed Alzheimer’s. “We looked at levels of these proteins in those cohorts in blood and found that the levels of those proteins were able to project risk of developing Alzheimer’s or dementia decades later down the road,” Johnson said. “Some of these modules are very likely to be causative and real risk factors for the disease,” — though the models would need confirmation from many additional studies before assuming clinical relevance.

New insights into how APOE4 drives Alzheimer’s risk would be extremely valuable, said Quinn. “Even though we’ve known about APOE4 being a risk factor for Alzheimer’s disease for 25+ years, we still haven’t really sorted out the mechanism by which it operates,” he said. “We have the means to identify people who are E4 carriers or E4 homozygotes, and we are still a little bit empty-handed in terms of things to offer those people.” The same biomarkers could potentially be used as outcome measures in clinical trials, Quinn added, if they reflect important disease mechanisms — though this application is “more of a stretch,” he said.

At a more basic level, maps of protein expression in Alzheimer’s patients could help researchers better capture the disease’s heterogeneity. “AD is not a pure disease, most often,” said Leslie Shaw, director of the biomarker research laboratory at the University of Pennsylvania, who was not involved in the research. Up to 50% of patients have pathology associated with Lewy body dementia, for example. “When you’re trying to predict how somebody is going to do in five years, you want to know not only their amyloid and their tau, you want to know their synuclein stats.”

By clustering people into groups based on similarity in proteomic features, Johnson and his colleagues identified 10 groups with distinct pathologies. We’re far from diagnosing Alzheimer’s subtypes with that degree of precision. But some experts are hopeful. “Longstanding speculation about the prospect of tailoring interventions to individuals based on personal subtype classification may be within reach,” said Gandy.

STAT encourages you to share your voice. We welcome your commentary, criticism, and expertise on our subscriber-only platform, STAT+ Connect

To submit a correction request, please visit our Contact Us page.