Skip to main content
Premium Trial:

Request an Annual Quote

Consortium Develops Algorithm from Blood-Based Alzheimer's Biomarkers

By Andrea Anderson

NEW YORK (GenomeWeb News) – In a study appearing in the newest issue of the Archives of Neurology, members of the Texas Alzheimer's Research Consortium reported that they have identified protein biomarkers in the blood that can be used to distinguish between individuals with and without Alzheimer's disease.

The researchers compared protein patterns in blood samples from hundreds of individuals with or without Alzheimer's disease and incorporated these potential biomarkers into an algorithm for detecting Alzheimer's cases in a test group. Their results so far suggest this algorithm can accurately classify most Alzheimer's cases — particularly when combined with APOE status and demographic data.

And, they said, information from the biomarker study is offering clues about possible sub-groups within Alzheimer's. For instance, by looking at some of the proteins that are frequently expressed at different levels in the blood of those with Alzheimer's disease, the team identified a potential subset of individuals with inflammation-related biomarker patterns.

"The obvious implications of any biological sub-group is targeted therapeutics and we're very excited about that," lead author Sid O'Bryant, a neurology researcher at Texas Tech University, told GenomeWeb Daily News. The team hopes to ultimately pursue studies exploring whether possible Alzheimer's disease sub-groups will respond differently to distinct therapeutic strategies, he explained.

Last summer, an international research group reported that they had identified biomarkers in cerebrospinal fluid that could be used to predict which individuals with mild cognitive impairment will go on to develop Alzheimer's disease.

But potential Alzheimer's disease biomarkers in the blood are also garnering attention since samples are easy to obtain and relatively non-invasive, O'Bryant and his co-authors explained. For instance, a 2007 study in Nature Genetics pinpointed 18 proteins in blood plasma that could be used to help distinguish between Alzheimer's cases and unaffected controls.

For the current study, done as part of a larger project involving hundreds of Alzheimer's cases and controls, members of the Texas Alzheimer's Research Consortium used the Austin-based biomarker company Rules Based Medicine's HumanMAP assay to look for Alzheimer's-related biomarkers in blood samples from individuals with Alzheimer's compared to healthy controls.

Using this multiplexed immunoassay human multianalyte profiling approach, the team randomly tested half of the 197 affected individuals and half of the 203 control individuals, identifying more than 100 proteins that were differentially expressed in the Alzheimer's group.

They then came up with an algorithm for discerning cases from controls based on this biomarker data. Indeed, the researchers noted, when they tried out the algorithm in a test group of affected and unaffected individuals from the other half of the study cohort, they found that the algorithm could accurately detect 80 percent of Alzheimer's cases with 91 percent specificity.

Adding information on APOE epsilon 4 status or demographic factors such as sex, age, or education further improved both the sensitivity and specificity of the algorithm, they noted, bumping the sensitivity up to 94 percent and the specificity to 84 percent.

When they looked more closely at more than two dozen of the most informative proteins expressed at higher or lower levels in the Alzheimer's patients, the team also found an over-representation of proteins involved in inflammation-related processes.

"We have a theory that there are endophenotypes in Alzheimer's disease, so that there are some sub-groups based on their biological profiles," O'Bryant explained.

"One of the sub-groups or endophenotypes that we're interested in is inflammatory," he added, noting that the current study — combined with follow-up studies — hint at an inflammatory endophenotype for Alzheimer's disease.

That, in turn, suggests it may eventually be possible to classify and treat Alzheimer's patients based on their endophenotype.

"[T]he identification of pathway-specific endophenotypes among patients with [Alzheimer's disease] would likewise have implications for targeted therapeutics as well as understanding differential progression among diagnosed cases," the team wrote.

Nevertheless, those involved cautioned that more research is needed to verify the study's findings and to determine whether the biomarkers used in the algorithm are specific to Alzheimer's or tied to conditions that may co-occur with it.

"Obviously the next step is to validate it in another set of Alzheimer's cases," O'Bryant said. "That's the next step we're planning right now."

Down the road, the researchers also hope to test the utility of the biomarker algorithm for distinguishing between different types of dementia and predicting Alzheimer's progression.

"I think that within a reasonable amount of time there will be a blood test for Alzheimer's disease out there that will allow for more rapid screening to be followed up by things like advanced neuroimaging," O'Bryant said.