NEW YORK – A team led by researchers at Emory University has identified a panel of cerebrospinal fluid proteins linked to Alzheimer's disease.
Detailed in a study published this month in Science Translational Medicine, the 48-protein panel could improve the performance of existing diagnostic markers for Alzheimer's as well as help predict progression of the disease, said Nicholas Seyfried, professor of biochemistry and neurology at Emory and an author on the paper.
In a cohort of 706 individuals participating in the Alzheimer's Disease Neuroimaging Initiative (ADNI), the protein panel exceeded the performance of the traditional Alzheimer's markers CSF amyloid beta (Aβ) 42, total tau, and phosphorylated-tau 181 for predicting clinical diagnosis of Alzheimer's, demonstrating an area under the curve of 0.94 compared to 0.90 for the traditional markers. The combination of traditional CSF markers and the protein panel showed the strongest performance for predicting clinical diagnosis of the condition, with an AUC of 0.96.
The protein panel also matched or exceeded the performance of traditional markers for predicting measures of neurodegeneration including fluorodeoxyglucose PET and hippocampal volume as well as several measures of cognitive decline and dementia.
The researchers also investigated whether the panel could help predict disease progression, looking at the trajectory of ADNI participants across a minimum of three visits over the course of at least three years. They found that combining the panel proteins with traditional CSF markers improved prediction of these patients' rate of cognitive decline, change in dementia severity, and change in hippocampal volume compared to using either set of markers alone.
Seyfried said this ability to predict progression is among the more significant contributions of the new marker panel.
"Existing markers do a great job of telling you whether someone has underlying Alzheimer's pathology," he said. "And they occur very early in the course of the disease. But they don't predict the onset of dementia, which is a different thing, because we know people have a high degree of amyloid and tau in their brains decades before they actually get cognitive impairment."
He suggested that the marker panel could help clinicians better predict if and when individuals in that preclinical phase will convert to being symptomatic.
He added that such predictions are likely to become more important given recent and anticipated Alzheimer's drug approvals.
"If you have biomarker panels that are predictive of the onset of cognitive decline … it could help decide whether someone is enrolled in a clinical trial or is given a drug," he said. "There are a lot of people positive for amyloid and p-tau, but if they are walking around healthy and normal, when do you decide to actually treat them?"
He said that because the panel looks beyond the amyloid pathology that has been the focus of much Alzheimer's drug and biomarker development work, it could prove helpful in evaluating the effectiveness of future drugs or drug candidates targeting other parts of the biology underlying the disease.
The protein panel emerged from previous discovery proteomics work by Seyfried and his colleagues in which they used mass spectrometry to measure proteins in both brain tissue and CSF from individuals with Alzheimer's and then conducted network analyses to identify CSF proteins linked to Alzheimer's pathology in the brain tissue.
That discovery effort yielded a variety of proteins linked to a variety of processes connected to Alzheimer's disease, including synaptic biology, metabolism, and amyloid biology. Taking those proteins, the Emory team then developed the smaller 48-protein panel that could be plausibly run in large numbers of samples, building selected reaction monitoring mass spec assays using isotopically labeled peptide standards for quantitation of each marker.
The researchers initially tested the panel in a set of roughly 400 samples from Emory. A team led by Erik Johnson, an Emory neurologist and coauthor on the STM paper, also used a version of the panel in an August Nature Medicine study looking at individuals with autosomal dominant Alzheimer's.
The Emory researchers have also used a version of the panel as part of research looking at differences in Alzheimer's pathology in Caucasian and African American individuals. The overwhelming majority of Alzheimer's biomarker development has been done in Caucasian cohorts, and studies have found that some blood-based biomarkers for Alzheimer's deliver different results in white and African American patients. Seyfried noted that he and his colleagues found that their panel's results differed by race, as well, with, for instance, some synaptic and metabolic proteins behaving differently.
Seyfried said that moving forward, the Emory team plans to test the panel in larger and more diverse sample cohorts. They are also interested in using it as part of clinical trials for Alzheimer's therapies being run at the university with a primary question being whether the markers are modifiable.
"The big question everyone asks us is, 'If they are predictive … could you modify them?'" he said. "Not even through drugs, per se, but through exercise and diet and other lifestyle interventions. Modifiable versions of these markers could be very informative."
Seyfried said the researchers also hope to develop a similar panel in plasma. Plasma markers are an area of great interest in Alzheimer's research and, increasingly, in clinical practice, as, unlike CSF, they don't require a lumbar puncture for sampling and are much less expensive than imaging.
He said that he and his colleagues will not look to simply move the CSF assay to plasma but would instead attempt to build a plasma panel much as they did the CSF panel by integrating discovery proteomic measurements in brain tissue with plasma analyses.
"I think it's going to be a similar work-up, and it is going to take a couple more years to figure out," he said. "In some of the work that has been coming out in plasma proteomics, the markers in the plasma associated with cognitive decline are different from those in the CSF. It will be interesting to see how that resolves over the next couple of years."