Despite extensive research on Alzheimer's disease (AD), this devastating neurodegenerative condition is an enormous continuing social and public health problem. More than 5 million Americans live with AD, and the total national cost of caring for people with AD and other dementias is projected to reach $214 billion during 2014, according to the Alzheimer's Association. The disease also takes a huge toll on the nation's 15.5 million caregivers—usually family members—who, as the disease progresses, provide nearly constant unpaid care. Without medical advances leading to viable treatments, the already substantial burden of AD is expected to continue to weigh down society and the healthcare system. Imagine three times more AD patients, a figure the Alzheimer's Association predicts could materialize by 2050 absent significant diagnostic and treatment strides.

The Food and Drug Administration has approved five drugs that temporarily offset memory and other cognitive loss and three agents for use with positron emission tomography (PET) of the brain. PET estimates amyloidβ (Aβ) plaque density in cognitively impaired patients who are being evaluated for AD and other causes of cognitive decline. But a more cost-effective test to spot early AD remains elusive.

Increasingly, research is focusing on prevention. "There's been a paradigm shift from trying to find a cure to stopping the disease from happening in the first place," noted Anne Fagan, PhD, professor of neurology at Washington University in St. Louis and a researcher in the Dominantly Inherited Alzheimer Network (DIAN). Reliable biomarkers are crucial to this strategy. Ideally, they not only would identify who will develop the condition and accurately measure its progression, but also enable researchers to test the efficacy of experimental treatments, she explained.

Studies have revealed several potential protein biomarkers in cerebrospinal fluid (CSF), with tau and Aβ among the most promising. Meanwhile, most research on proteins and other types of biomarkers in the blood, which are easier and cheaper to collect than CSF, has not been able to detect preclinical disease with high sensitivity and specificity.

However, a small proof-of-concept study from a team led by Howard Federoff, MD, PhD, executive dean at Georgetown University School of Medicine in Washington, D.C., showed that mass spectrometry analysis of 10 blood lipids distinguished, with 90% accuracy, who among an initial group of 525 cognitively healthy adults age 70 and older developed cognitive impairment over a 2–3 year period (Nature Medicine 2014; doi:10.1038/nm.3466). If these results are validated in larger and more involved studies, this line of research might someday lead to a relatively noninvasive test that could help clinicians determine who is at risk for the disease.

Finding the Lipids

Federoff's team enrolled mostly educated, middle-class Caucasians living in or near Rochester, N.Y. The researchers took blood samples and administered cognitive tests at baseline and at yearly follow-up visits. Over the course of the study, 28 participants developed either amnestic mild cognitive impairment (MCI) or AD.

The authors selected 18 of these patients and analyzed their plasma via mass spectrometry for lipid markers that differentiated them from 53 age, sex, and education-matched controls who did not have cognitive decline. Broad analysis of metabolites showed that levels of several plasma phospholipids important to the integrity and function of cell membranes differed between the groups. Follow-up analyses revealed eight phosphatidylcholines and two acylcarnitines as the most discriminating. Subjects who remained healthy had higher baseline levels of all 10, compared to those who developed MCI or AD.

In contrast to extensive research that has shown the genetic variant ApoE4 as a strong risk factor for AD, this study found no relationship between this known risk factor and the lipids in the biosignature. That's partially because it was designed to examine metabolites and not the gene variant as a primary outcome, said the paper's first author, Mark Mapstone, PhD, associate professor of neurology at University of Rochester School of Medicine. Many researchers who work with lipids try using ApoE4 to make panels more predictive, so editors reviewing the paper "thought we should look at APOE because it's a strong risk factor," he explained. "But the classification of the subjects based on the biosignature was not related to ApoE4 presence."

AD researcher Ling Li, PhD, a professor of experimental and clinical pharmacology at the University of Minnesota College of Pharmacy, said one reason ApoE4 did not prove to be predictive might be the study's small sample size.

Does Cholesterol Play a Role in Alzheimer’s Disease?

The Crucial Validation Step

Validation of the Federoff team's findings is crucial, said Mapstone, who offered a glimpse into their plans to do so. "The first step would be to validate the panel in another, more ethnically diverse population. Ours was 98% Caucasian. Probably we'd also try to get younger people, maybe in their 60s and 70s, to push back the range a little," he explained. Mapstone's co-investigator, Amrita Cheema, PhD, an associate professor and co-director of the Proteomics and Metabolomics Shared Resource at Georgetown University will discuss the team's findings in a late-breaking session at the 2014 AACC Annual Meeting & Clinical Lab Expo.

Researchers not involved in this study offered more ideas for validating the 10 analytes. Li suggested both cross-sectional and longitudinal studies because AD takes decades to develop. "It would be good to correlate the panel with other biomarkers already out there, like amyloidβ and tau, and with amyloid imaging," she said, also recommending study of the 10-lipid panel in patients with the inherited form of the disease.

Fagan cited interest the research community would have in knowing more details about the clinical characteristics of any follow-up study cohorts, and in seeing future studies use more standard definitions of normal cognition and what constitutes disease. She also observed the necessity of including controls "with no underlying Alzheimer pathology. If there's not a well-defined cohort, it's hard to interpret the data."

Validation studies should also pay careful attention to how samples are prepared and the many factors that can affect measured blood lipid levels, including if patients fasted, the time of day, and choice of blood collection tubes, noted Leslie Shaw, PhD, who runs the Alzheimer's Disease Neuroimaging Initiative (ADNI) biomarker core and is a professor of pathology and laboratory medicine at University of Pennsylvania Perelman School of Medicine in Philadelphia.

Keith Fargo, PhD, director of scientific programs and outreach for the Alzheimer's Association, said that the 10-lipid panel and any other useful biomarker would need to show evidence that it is predictive "in thousands of people" and indicate changes in the brain in response to treatment.

If properly validated and implemented, the lipid test could pose technical challenges for some labs, Li pointed out. "Measuring these lipids is not like doing routine blood cholesterol and lipid panels, which can be accomplished by easy biochemical assays." In part that's because this specific lipid panel currently relies on sophisticated preparation and mass spectrometry, which many labs lack. "But there's big potential in the test because it uses blood," she emphasized.

Other Recent Biomarker Evidence

Most studies of lipid and other biomarkers have focused on CSF in symptomatic patients. The few that have examined blood lipids in preclinical disease have not yielded markers as accurate as those described by the Georgetown-led team, said Mapstone. However, a study of several proteins in the blood came fairly close, but only in people with diagnosed AD. This study showed that 18 proteins tested by the Australian Imaging Biomarker and Lifestyle Research Group and validated in an ADNI cohort distinguished individuals with AD from cognitively healthy control subjects with a mean sensitivity of 85% and specificity of 93%. Of the 18 proteins, 11 were significantly increased in AD, while seven were significantly decreased (Arch Neurol 2012;69:1318–25).

Meanwhile, a more recent study suggested that lipid metabolites could advance understanding of early disease mechanisms and identified some of the same metabolites as the Federoff team's study. Mass spectrometry of CSF and plasma samples from 45 people in the Mayo Clinic Study on Aging and Mayo Clinic Alzheimer's Disease Center revealed that those with MCI and AD had significant changes in CSF and plasma, while changes in plasma accurately reflected changes in CSF. Overall, the researchers identified 342 and 351 significantly altered plasma and CSF metabolites, respectively (PLoS One 2013; doi:10.1371/journal.pone.0063644).

Tau, a well-established biomarker in Alzheimer's studies, thus far has been reliably detectable only in CSF, but another recent study indicates headway in detecting the protein in circulating blood via an assay that employs arrays. A cross-sectional study of 54 patients with AD, 75 with MCI, and 25 cognitively normal controls found plasma tau concentrations to be significantly higher in AD patients compared with both controls and MCI patients. However, researchers found no correlation between tau levels in plasma and CSF (Alzheimers Res Ther 2013;5:9).

What's the Change Over Time?

Meanwhile, biomarker data from patients with inherited disease in another study that included both cross-sectional and longitudinal data analysis emphasized the importance of looking at changes within individuals as part of the search for predictive biomarkers (Sci Transl Med 2014;6:226ra30).

Although autosomal dominant disease caused by mutations in three genes comprises < 5% of all AD cases, DIAN and other research efforts focused on families with these mutations to help characterize biomarker changes across the disease process. That's not only because dementia is a certainty in people with the mutations, but also "because they develop symptoms at an age similar to their parents so we have a better idea where they fall in the course of their disease," explained Fagan. She was lead author of the Science Translational Medicine paper, which confirmed prior findings about biomarker changes in inherited AD and raised questions about how a few analytes behave over the course of the disease.

Specifically, Fagan's study confirmed previous DIAN cross-sectional findings that indicate AD pathology exists 10–15 years before symptom onset. The initial findings came from studying 79 mutation carriers' plasma Aβ1-42 and CSF Aβ1-42 and tau in relation to their individual estimated age at symptom onset (N Engl J Med 2012;367:795–804).

The cross-sectional part of Fagan's study included evaluation of baseline data for five plasma and five CSF biomarkers from a larger DIAN cohort, including 96 non-carriers and 146 participants with a variety of mutations in PSEN1, PSEN2, and APP. Mutation carriers had amyloid imaging and plasma analysis for four types of Aβ fragments, and visinin-like protein-1 (VILIP-1), a neuron-specific intracellular calcium sensor protein, a proposed marker of neuronal cell injury and death. The researchers also analyzed CSF markers including Aβ1-40, Aβ1-42, tau, phosphorylated tau (p-tau), and VILIP-1.

In this study, reduced concentrations of CSF Aβ1-42 were associated with the presence of amyloid plaque in the brain as evidenced by amyloid imaging, similar to what has been shown in late-onset, sporadic AD. The researchers saw higher-than-normal levels of CSF Aβ1-42 concentrations in carriers at least 25 years before their expected age of symptom onset. However, while controls' Aβ1-42 concentrations appeared to remain constant, carriers' concentrations were significantly lower than controls' values at least 10 years before their estimated age of symptom onset. Meanwhile, tau, p-tau, and VILIP-1 concentrations were elevated in carriers 15–20 years prior to symptoms, and remained higher than controls' at subsequent ages, similar to what is observed in the more common late-onset form of AD.

In the within-individual longitudinal analysis involving 11 noncarriers and 26 carriers, CSF tau also rose over a period of 1–3 years in carriers prior to their expected age of symptom onset. But in carriers who were symptomatic and past their expected ages of onset, levels of tau, p-tau, and VILIP-1 all fell slightly, by about 10 pg/mL, over time. Fagan speculated that drop in these analytes may reflect slowing release of these proteins from the brain into CSF as the rate of acute neuron cell injury and death diminishes and tissue shrinks.

Like the Federoff team's study, these findings need to be further validated. If confirmed, they might warrant adjustment of the current thinking about AD's biomarker trajectories and could affect evaluation of particular biomarkers' behavior in clinical trials of disease modifying therapies, said Fagan. "These data also highlight that we should not make assumptions about biomarker changes over time from cross-sectional studies," Fagan added. ADNI researchers observed a similar drop in tau over time among some ADNI participants with sporadic disease (Acta Neuropathol 2013;126:659–70).

Eventually, Fagan's findings and other studies of inherited disease may inform the more common sporadic form of AD, according to Shaw. Comparing and contrasting biomarker findings in studies of patients with inherited and sporadic disease will reveal commonalities in what's behind cognitive decline in the two populations, he added. Studying inherited disease may also inform understanding of sporadic disease because familial AD patients are younger, and therefore less likely to have signs of other brain pathologies like vascular disease, which is common in older patients with sporadic AD.

The field "is getting closer and closer to viable biomarker tests," Fargo noted. In addition to studies examining how biomarkers show disease progression, he pointed to progress on other fronts such as standardizing reagents and fluid handling techniques. Mapstone is confident that research will yield viable biomarkers, and urged laboratorians to prepare themselves. "Clinical chemists will implement our findings and make them into CLIA-certified tests we can use to help patients," he said. "You will be called upon to assist us."

Deborah Levenson is a freelance writer in College Park, Md. She can be reached at