A consortium of researchers working as part of the Alzheimer's Disease Neuroimaging Initiative has completed an analysis of potential plasma protein biomarkers for Alzheimer's in three separate patient cohorts, identifying four analytes associated with the disease.
Detailed in a paper published this month in Neurology, the study identified the protein apoE, B-type natriuretic protein, C-reactive protein, and pancreatic polypeptide as linked to very mild cognitive dementia/mild cognitive impairment/Alzheimer's across all three cohorts.
Using Rules-Based Medicine's (now Myriad RBM) Human DiscoveryMAP immunoassay platform, the researchers measured levels of 190 proteins in two independently collected cohorts – one a 333-patient cohort collected by clinicians at Washington University in St. Louis and the other a 267-patient cohort collected by researchers at the University of Pennsylvania.
A comparison of these two cohorts identified 17 analytes associated with very mild dementia/MCI/Alzheimer's. The researchers then validated these findings via comparison with data from an independent 566-patient ADNI cohort run on the same Human DiscoveryMAP platform, identifying the four potential markers.
In an e-mail to ProteoMonitor this week, Alzheimer's researcher Hugo Vanderstichele, who was not involved in the research, said the study was "well-designed," noting as a particular advantage "that the results [were] confirmed in different sample sets, not always collected and stored under identical conditions."
The founder of Alzheimer's biomarker firm ADx Neurosciences and formerly a developer of commercial assays for cerebrospinal fluid-based tests for the gold-standard Alzheimer biomarkers tau, p-tau, and Aβ42 at Innogenetics – now part of Fujirebio – Vanderstichele added that while "more validation studies will be required ... the initial outcome of the study is promising."
Much Alzheimer's biomarker work has focused on CSF markers like tau and Aβ42. However, sample collection for CSF studies and clinical applications requires patients to undergo a lumbar puncture, making plasma markers desirable due to the relative ease of obtaining samples for analysis.
Alzheimer's markers have typically been more difficult to detect in plasma than CSF, however, and some high-profile Alzheimer's plasma biomarker research has shown poor reproducibility (PM 1/30/2012).
As Vanderstichele noted, the ADNI study is promising given its use of three large, independent patient cohorts, a feature that sets it apart from many previous Alzheimer's plasma biomarker discovery efforts, which have often relied on single, smaller sample sets.
"I think the biggest problem in discovery science ... is the failure to replicate across cohorts," William Hu, a researcher at Emory University and lead author on the Neurology paper, told ProteoMonitor.
"We've been looking at this issue quite a bit first in terms of cerebrospinal fluid studies and now in plasma studies, and obviously plasma protein changes are subject to many factors other than Alzheimer's disease," Hu added. "If you have a group of Alzheimer's patients who also have high blood pressure and high cholesterol and you have a group of controls that are super healthy, any plasma signal you get can be due to a non-Alzheimer's disease factor, so we really wanted to replicate the results across cohorts."
Vanderstichele suggested, though, that making comparisons across independent sample cohorts comes with its own challenges, particularly in the case of Alzheimer's disease where diagnosis and distinguishing between stages of the disease typically relies on a clinical assessment.
"The comparison of results is linked to the selection of the patients in the study," he said. "It would have been good to see a comparison of the plasma markers using subjects with a biomarker profile identified as healthy or [Alzheimer's] using CSF biomarkers."
Hu agreed that establishing a diagnostic gold standard across the three groups was a challenge, adding that because of this the study asserted only that the four proteins were associated with very mild cognitive dementia/MCI/Alzheimer's broadly and didn't make claims as to which stages of the disease they might be linked to.
"Every center uses slightly different terminology [for classifying Alzheimer's], and even within the same terminology there's a lot of subjective clinician input," he said. "We didn't do classification because we didn't have a gold standard, but what we do know is that when we look at a population of cognitively impaired, the majority of which – over 80 percent – have Alzheimer's disease, and a group of controls ... the big signal changes we found across all three cohorts were those four. So even with the clinical diagnostic standardization aside, I think we can find a reliable signal that's associated with [Alzheimer's]."
Hu noted that the study did compare the markers to CSF biomarker data in the case of the ADNI cohort, finding, he said, that "the same signals were identified, which certainly added to our confidence level." Suitable CSF samples were not available for all of the patients in the Washington University and Penn cohorts, he said.
Alzheimer's biomarkers are a significant area of focus for proteomic research, with early detection and progression biomarkers, in particular, seen as key to pharmaceutical research into drugs for the disease. In addition to the ADNI work, a number of researchers outside the consortium are pursuing plasma and CSF markers for the disease, as well.
In March, for instance, Proteome Sciences, the UK's National Institute for Health Research, and Millipore completed a 1,000-sample validation study of blood-based Alzheimer's markers, identifying three separate panels – each containing between 11 and 16 proteins – that can distinguish between controls, mild cognitive impairment, and Alzheimer's (PM 3/30/2012).
Hu said the plasma markers identified in the Neurology study would likely be most useful as early detection markers, noting that "there is a lot more interest for progression [markers in CSF] than blood simply because diagnostically blood is not quite there yet."
"Blood will probably be more or less a screening tool for [moving patients to the] next stage of workup to include, potentially, spinal fluid analysis," he added.
Moving forward, the researchers hope to further validate the identified markers in a North American or international consortium, Hu said. He noted that there was interest in such an idea at the Alzheimer's Association International Conference last month in Vancouver, and added that the community was looking to hold another meeting this fall to discuss how to go about additional work on the markers.
Hu said the researchers might also reexamine certain markers that they had been able to replicate in two of the three cohorts. "Just because one marker doesn't replicate in the other cohort doesn't mean it's not real," he said. "There were markers [for instance] that changed in Penn and changed in ADNI, but didn't change in Wash U. Those things are worth a second look."
He stressed, though, the importance of replicating results across centers and cohorts, noting that "one contribution we made is the identification of these markers, but a more important one perhaps is to lay the groundwork for other people who want to do discovery work ... [showing] that you really need to replicate not only across platforms but across centers and populations, so that any marker you identify in Atlanta, for example, will also be applicable in Seattle."
"If that doesn't work then it's not a reliable biomarker," he said.