NEW YORK (GenomeWeb) – A mass spectrometry-based analysis of cerebrospinal fluid in Alzheimer's disease patients has yielded a handful of potential biomarkers for the disease.
Led by first author Payam Khoonsari and Kim Kultima of Uppsala University, and co-senior author Ganna Shevchenko of the KTH Royal Institute of Technology, the researchers used label-free shotgun mass spectrometry to look at proteins in the cerebrospinal fluid of 10 Alzheimer's disease patients and 10 healthy controls. They also performed protein depletion of high-abundance proteins to improve detection and quantification of low-abundance proteins.
As they reported Monday in PLOS ONE, the authors found eight proteins that were differentially expressed between the two study groups. "ApoM, LRG, FBLN3, and PTPRZ have functions related to cell adhesion, migration, and morphology," and may also be associated with other aging-associated diseases like cancer and diabetes, they wrote. C1QB, C1QC, complement C1S, and SEZ6 may be implicated in synapse development.
"Cerebrospinal fluid is a proximal fluid in direct contact with the brain interstitial fluid that potentially reflects biochemical changes related to [the] central nervous system, making it a promising source of biomarkers in neurological disorders such as AD," they added. While Alzheimer's disease is associated with several proteomic markers, especially the protein tau and beta-amyloid peptides, those have limited value for monitoring disease progression.
"There are a large number of proteins proposed to be specifically related to AD but many of those have not been replicated in the similar studies, making them unreliable to be used in the clinic," Kultima told GenomeWeb in an email. "This low reproducibility can be because of multiple factors such as differences in the cohorts, non-identical experimental workflows, as well as different methods of data pre-processing and downstream analysis, and also making wrong assumptions about the data."
He added that although this study started with the aim of finding potential Alzheimer's biomarkers, "we also evaluated a number of the most common data pre-processing programs and statistical assumptions about the data."
The researchers performed trypsin digestion and nanospray liquid chromatography tandem mass spectrometry (nanoLC-MS/MS), and then used antibody suspension bead arrays to validate the mass spectrometry results.
They then used no less than five different processing programs — PEAKS, Maxquant, OpenMS, DecyderMS, and Sieve — to analyze the results. The programs identified a total of 894 different proteins in the cerebrospinal fluid, 173 of which were identified by all five programs. With further analysis, the researchers found 162 proteins to be significantly altered in the Alzheimer's group, 31 of which were found to be statistically significant by all five programs.
"There was a large difference between the packages in terms of identification," Kultima said, adding that while it's normal for each program to miss some proteins, it's a different matter when they're identified but not considered a high-quality target. "It is especially important to extract as much as information from the software packages as possible and make it publicly available so it can be used by other researchers to compare reproducibility of their results," he noted.
After running the data through the programs, the team then normalized the information with spiked-in chicken ovalbumin, and compared the mass spectrometry analysis with the antibody analysis. This led them to narrow the list of potential biomarkers to eight proteins, all of which displayed lower levels in Alzheimer's patients than in the controls.
"By performing local normalization based on a spiked-in protein, thus only correcting for experimental bias, the correlations to the antibody-based profiling results were substantially increased," the authors wrote. "We suggest adding one or several recombinant proteins from a different species than the investigated, which can be used for correcting for experimental bias and to investigate if the assumption for using global normalization is valid."
"By performing any kind of statistical analysis, some assumption are being made about the data," Kultima explained to GenomeWeb. "Many people tend to use the methods commonly employed by other research groups without evaluating the assumptions. This is especially important when the data is to be normalized to correct for systematic bias. An improper method can, in practice, cause more bias and unwanted variation and can severely affect the biological conclusion."
In their paper, the authors also pointed out other factors that might have influenced results. In addition to the small study size, they noted that there was a slight age different between the two groups, where the controls were, on average, nine years older than participants in the Alzheimer's group — protein levels in cerebrospinal fluid are thought to change with age. The lower protein levels could also be the effect of protein depletion.
The scientists said they are conducting a larger study to further validate the protein biomarkers.