NEW YORK – New research has uncovered protein signatures associated with different forms of Alzheimer's disease (AD), including markers found in plasma, cerebrospinal fluid (CSF), and brain tissue.
The study, led by a team at Washington University School of Medicine, appeared in Science Translational Medicine on Wednesday.
"We were able to leverage those proteins to generate new prediction models for AD," senior and corresponding author Carlos Cruchaga, a researcher at WashU, said in an email.
Using aptamer technology from SomaLogic, the researchers assessed levels of more than 1,300 proteins in brain, CSF, and plasma samples from individuals with or without AD, including sporadic AD cases; individuals with an autosomal dominant form of AD linked to variants in presenilin-1, presenilin-2, or amyloid-related genes; and cases in individuals carrying risk variants in the TREM2 gene.
After quality control, data processing, and other steps, the team was left with quantitative data for 1,092 proteins in postmortem brain samples from 290 cases of general AD, two dozen cases of autosomal dominant AD, 21 TREM2 risk variant carriers, and 25 cognitively unaffected control individuals.
In CSF samples from 176 AD cases, 47 individuals with TREM2 risk variants, and 494 controls, meanwhile, the investigators quantified 713 proteins. They also assessed plasma proteomic profiles spanning 931 proteins for 105 AD cases, 131 TREM2 risk variant carriers, and 254 healthy control individuals.
After replication testing, the team tracked down 40 proteins with altered representation in the CSF samples from individuals with sporadic AD, along with nine potential plasma protein markers and eight brain protein markers. Similar shifts turned up in samples from those with autosomal dominant forms of AD, though the effects of these changes appeared to be more pronounced.
"For autosomal dominant AD, we found a subset of proteins that were unique to this population, but others overlapped with sporadic AD," Cruchaga said, noting that the overlapping protein set had a larger effect size in autosomal AD "indicating the disease is more severe in this group."
The researchers went on to highlight distinct sets of proteins with altered representation in the brain tissue, CSF, or plasma samples from TREM2 carriers, he added, including an enrichment for proteins with ties to innate immune system processes.
At least some of the brain proteins linked to autosomal dominant AD were also found at altered levels in blood or CSF samples, as well as in existing mass spec data for AD patients from several large studies, the team reported.
"The replicated proteins were used to create brain-, CSF-, and plasma-specific prediction models and to identify pathways leading to disease," the authors explained, noting that they developed an interactive web portal for visualizing and analyzing data from the current study.
In particular, the researchers reported that a proteomics-based AD risk model based on CSF data appeared to outperform existing phospho-tau/amyloid beta (AB) models for predicting sporadic forms of the disease, while a blood-based model showed prediction results resembling those of the existing CSF model.
"These prediction models, specially the one in plasma, is highly relevant to the field [as it is] the first non-AB and tau prediction model," Cruchaga said.
In contrast to conventional CSF AB/ptau models that tend to show reduced performance for predicting AD in TREM2 risk variant carriers, he explained, the proteomic approach "was able to identify a subset of proteins that can identify TREM2 risk variant AD cases from controls and sporadic AD."
When the researchers looked at the pathways that the potential protein markers belonged to, they saw an overrepresentation of calcineurin and Apo E pathways previously linked to AD, along with pathways that are more typically associated with Parkinson's disease or innate immune function, including the alpha-synuclein and LRRK2 pathways.
"Although additional validation of some of our findings will be needed, these results highlight the need to combine brain tissue, CSF, and plasma proteomics to fully understand the biology of AD and to create prediction models for individuals with AD with specific genetic profiles," the authors concluded.