Alzheimer's Early-Warning Biomarkers: Are We There Yet?

by Jim Schnabel

April 8, 2014

Arguably the weapon most needed now in the fight against Alzheimer's is not a new drug but a new test-a test to predict the onset of dementia years before it occurs.

Such a test would make use of "biomarkers" of pre-symptomatic Alzheimer's in blood, cerebrospinal fluid and/or brain scans. Initially it would be useful in selecting and monitoring subjects in clinical trials of preventive drugs, and in principle could make such trials faster and cheaper to conduct.

Ultimately, if any drug proves useful in preventing or delaying Alzheimer's dementia, an early warning test could allow people to get that drug in time. The many failed clinical trials of recent years suggest that candidate Alzheimer's drugs so far have been delivered too late in the disease course, when extensive neuronal losses have already occurred and the driving processes of degeneration have become complex and far advanced.

Many laboratories around the world have been working to develop biomarker-based early warning tests for Alzheimer's, and there are signs of significant progress. "We're definitely getting closer to bringing biomarkers to the clinic," says Anne Fagan, an Alzheimer's researcher at Washington University-St. Louis.

A blood-based marker of neuronal destruction?

One hopeful report appeared in Nature Medicine in March: A team led by Howard Federoff at Georgetown University School of Medicine, and Mark Mapstone of the University of Rochester School of Medicine, discovered an apparently useful early-warning biomarker based on bloodstream levels of ten small "phospholipid" molecules.

To find the biomarker, the researchers enrolled 525 people who were aged 70 and over. Some of the subjects were diagnosed at entry with Alzheimer's or a condition that often precedes Alzheimer's known as amnestic mild cognitive impairment (aMCI). Other subjects tested cognitively normal at entry but converted to aMCI or Alzheimer's during the five-year study. The rest stayed cognitively normal throughout the study.

Federoff and his colleagues measured levels of a long list of (plausibly Alzheimer's-related) small molecules, also known as metabolites, in blood samples from the subjects. They looked for those that differed the most between cognitively normal subjects that didn't develop Alzheimer's or aMCI during the study, and those that did.

The metabolite measurements revealed thousands of molecules-including fat-related lipids, amino acids and neurotransmitters-whose levels tended to move up or down during the transition from cognitive normalcy to aMCI or Alzheimer's. The researchers analyzed these and came up with a high-reliability biomarker panel consisting of ten metabolites related to phosphatidylcholine (PC, also called lecithin) and acylcarnitine, both known as major components of neuronal membranes. The levels of all ten molecules tended to be lower in the just-before-disease group than in those who still had at least five years of cognitive normalcy ahead.

Using this collective decline in the ten phospholipids as the biomarker, the researchers applied it to a set of study subjects whose blood samples hadn't been used in the initial analysis. The new biomarker was able to classify, with 90% accuracy, both the pre-disease subjects and the staying-normals.

Federoff and colleagues suspect that the decline in the ten lipids, in the pre-disease cases, signifies the cascading destruction of brain cells in key regions that corresponds to the shift from subtle "presymptomatic" cognitive deficits to obvious memory loss and dementia.

One of the next steps is to further validate the biomarker array-and perhaps change and optimize it-through studies in new groups of patients. "We plan to clinically validate our findings using larger and more diverse cohorts," says Amrita Cheema, a metabolomics expert at Georgetown and a co-author of the study.

Such studies should clarify how specific this putative biomarker is for Alzheimer's/MCI. In principle, other neurodegenerative conditions, from Parkinson's and Huntington's diseases to fronto-temporal dementia, could cause the widespread loss of brain cells and associated drops in membrane-related lipids. "We do plan to characterize this panel to see if can be used as a generic determinant of cognitive decline or for ruling out other neurological disorders," says Cheema.

In any case, the researchers seem confident that their biomarker array will be useful in finding high-risk, pre-symptomatic subjects for clinical trials of candidate Alzheimer's drugs. "We'd like to use this as a screening measure to develop a trial for what we hope are disease-modifying therapies," says Mapstone. "If we can just give [the therapies] earlier, maybe they'd have a fighting chance."

The news of the biomarker panel results is welcome, says Fagan, but further studies are definitely needed: "I'm really for the idea of a plasma biomarker, but over the past ten years so many have come and gone and haven't been replicated."

For example, Stanford University researchers reported in 2007 that they had found an 18-protein biomarker panel that could identify pre-Alzheimer's cases, but Swedish researchers later were unable to reproduce the results in a different sample.

Tau and amyloid beta

A useful early-warning biomarker of Alzheimer's would be anything that changes reliably during the transition from normal cognition to MCI to dementia. Other biomarkers that have been studied include glucose uptake in Alzheimer's-prone brain regions; the magnetic resonance imaging (MRI) measured volume of the memory-related hippocampus region of the brain, which tends to shrink with dementia onset; and apolipoprotein-E, one variant of which (E4) brings a higher Alzheimer's risk and faster disease course. And, of course, much of the research on early-warning biomarkers for Alzheimer's has focused on two molecules that have always been linked to the disease process: tau and amyloid beta.

Aggregates of these proteins-from small and insoluble oligomers to the large and insoluble, plaque-making fibrils-accumulate in affected brain areas before and during the cognitive slide into dementia. Researchers so far have failed to find robust Alzheimer's-predictive shifts in the bloodstream levels of these proteins. However, measurements of their levels in cerebrospinal fluid (CSF)-sampled by spinal tap-have been more promising, to the extent that they are now beginning to be used in clinical trials as potential indicators of disease risk and disease progression.

CSF levels of amyloid beta, in particular the more aggregation-prone form known as Aβ42, tend to decline noticeably a decade or more before Alzheimer's dementia sets in. By contrast, CSF levels of total tau, and of the disease-linked, highly phosphorylated form called phospho-tau, rise in the decade before dementia.

Most of the conclusions so far about CSF tau and Aβ42 have been drawn from cross-sectional studies-each of which is a snapshot of a group of people with various ages and disease stages. But the clinical goal is to be able to track these markers in each person from middle age onwards, to determine Alzheimer's risk, somewhat as doctors now track cholesterol levels and blood pressure to determine heart attack and stroke risk. The needed "longitudinal" studies of CSF tau and Aβ42 levels in individuals over long periods, from cognitive normalcy to dementia, have only just begun to be published-Fagan and colleagues reported one of the first in March.

The amyloid-beta plaque level in the brain, as measured using plaque-marking positron emission tomography (PET) tracers, is another apparently useful indicator of Alzheimer's risk. Although plaque load doesn't correlate very closely with neuron loss and cognitive decline-generally some plaques are found even in cognitively healthy elderly brains at autopsy-a high plaque burden certainly is consistent with dementia. Doctors already are using amyloid imaging as a diagnostic aid-often to help differentiate Alzheimer's from other forms of dementia-in private clinics worldwide. A just-started Alzheimer's prevention trial, in which cognitively healthy elderly people will receive doses of anti-Aβ42 antibodies (or placebo) for three years, is enrolling subjects that show high plaque loads on amyloid imaging. By 2020, tau PET imaging technology also may be available.

Researchers and regulators still have to agree on standards for interpreting these biomarkers, how useful they really can be at predicting dementia in the clinic, and what comprises the most useful set of markers-a set that may include molecules, such as amyloid-beta and tau oligomers, that have not yet been studied widely as early-warning biomarkers to date. But Fagan envisions a time, not too many years hence, when pre-Alzheimer's tests will become almost as routine in middle age as other medical screening tests are today. "You might have an amyloid scan with a CSF tau with a lipid biomarker with a genetic test of apolipoprotein-E, that sort of thing," she says. "I think that's ultimately where it's going to go."