13 September 2008. Among the plethora of factors that might serve as useful biomarkers for Alzheimer disease, perhaps none draws as much scrutiny as plasma amyloid-β. Over the years, many studies have alternately supported and questioned the idea that Aβ in the blood reflects AD, or Aβ in the brain. Now, longitudinal data from one of the largest studies on plasma Aβ to date seems to shore up the value of the measurement. “We showed in the study that people who have high levels of plasma Aβ42, several years—and in some instances four to six years—before onset are at high risk of subsequently developing AD,” Richard Mayeux, Columbia University, and senior investigator on the study, told ARF. The work appears in this week’s PNAS online. Mayeux stressed that the finding does not translate into a blood test to predict AD. “We are not promoting this as a test by any means. I view this like cholesterol. Having high cholesterol doesn’t mean you are definitely going to have a heart attack or a stroke, but it does increase your risk,” he said.
The study enrolled more than 1,100 older adults (age 70 and up) living in northern Manhattan. Participants were clinically screened and put through a comprehensive neuropsychological test battery to ensure they were cognitively normal at study entry, and then followed every 18-24 months for up to six years. During this time, 104 (9.2 percent) developed AD. Plasma Aβ40 and Aβ42 were measured at baseline and after the second follow-up, i.e., after four to five years.
“This is an astonishing tour de force to have done so many samples, and they are fairly well characterized by this group,” said John Trojanowski, University of Pennsylvania. Trojanowski is the principal investigator for the biomarker core of the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Anne Fagan (no relation to this reporter), Biomarker Core leader for the Washington University Alzheimer’s Disease Research Center in St. Louis, Missouri, agreed with the assessment on sample size. “This is one of the great strengths of this study, but it could also be a weakness, too, because it becomes more difficult to control variables,” she said.
When analyzing the data, first author Nicole Schupf and colleagues found that people who were in the highest quartile for baseline plasma Aβ42 were about 3.5 times more likely to develop AD, suggesting increased risk. No such propensity was found for Aβ40. When the researchers analyzed the samples taken at the later time point, they found that those people whose plasma Aβ42 concentration had dropped since their baseline value were also more likely to convert to AD. “This was what we anticipated because you see this in transgenic animals,” said Mayeux. In transgenic mice that overproduce Aβ, levels of plasma, cerebrospinal fluid (CSF), and brain Aβ are low initially. Later, before the animals start to develop plaques, blood levels of Aβ are elevated compared to wild-type littermates; and later still, as the mice start to deposit plaques, their CSF and plasma Aβ42 levels decrease, Mayeux explained.
“The fact that the high Aβ people who went on to [develop] Alzheimer’s had a reduction in plasma Aβ is very interesting from the point of view of Aβ dynamics,” said Trojanowski. “It is important to have more understanding of Aβ in these fluids and of the dynamic changes that occur,” he said. Recently, for example, researchers led by David Holtzman at Washington University, St. Louis, Missouri, reported that the relationship between CSF Aβ, interstitial Aβ in the brain, and disease may be more complex than previously appreciated (see ARF related news story). Understanding Aβ dynamics may help researchers identify changes that are better diagnostic markers for disease.
One aspect of Aβ dynamics that directly affects levels of soluble Aβ is fibrillization, but most of the assays used to detect Aβ in the plasma are not capable of identifying fibrillar forms of the protein. To address this, Mayeux collaborated with Jeffrey Ravetch at Rockefeller University, also in New York. Ravetch’s group developed a monoclonal antibody, 13C3, that specifically recognizes protofibrillar forms of Aβ42. The researchers used this antibody to assay Aβ in plasma of a subset of 402 subjects. Protofibrillar Aβ was detected in 34 percent of this cohort and strongly correlated with plasma Aβ42 levels. Again levels of protofibrillar Aβ at follow-up significantly declined in people who went on to develop mild AD. Mayeux said that the origin of blood protofibrillar Aβ is uncertain. It is not clear if it forms inside or outside the brain. Though there is prior evidence for Aβ deposits in the skin, it is not clear where that Aβ comes from, either (see Joachim et al., 1989).
This study will no doubt rekindle the controversy about the value of plasma Aβ measurements. Many studies have failed to find any relationship between AD and plasma Aβ, and it is unclear why the data has been so equivocal (see ARF related news story). Mayeux stressed that timing is a critical factor. If samples are taken during MCI or just prior to onset of AD, then that underlying drop in plasma Aβ that the New York researchers detected might skew trends in the data.
Trojanowski and Fagan offered other thoughts on this critical issue. “Why the outcomes from other studies vary is not clear, but presumably there may be technical details that are very important,” suggested Trojanowski. Such nitty-gritty could include apparently mundane factors like the type of tubes used to collect samples or more obvious components such as antibodies and platforms used for ELISA tests, or diurnal fluctuations in plasma Aβ, he suggested. Fagan agreed. “We go to great pains to ensure our samples are collected in the same way at the same time of day,” she said.
Both Trojanowski and Fagan noted that lack of standardization is a huge problem with which the field is grappling. Many laboratories have begun to use commercially developed tests and platforms for detecting Aβ; however, at $100 or more per sample, these kits can be prohibitively expensive when studying thousands of patients, and many labs opt for “home brew” or in-house ELISA measurement. But even when different labs use the same protocol, the data can be confusing. “Consistently, people show that in CSF tau is up and Aβ is down in Alzheimer patients, but if you look at the values published from different centers, all using Innogenetic kits, there is two- to threefold variation in the levels in Aβ in patients, which reflects lack of standardization,” said Trojanowski. This is one issue ADNI hopes to address.
Going forward, Mayeux and colleagues plan to keep following this cohort with further time points of plasma Aβ analysis to get a higher-resolution picture of plasma Aβ dynamics. As for how this data might be used, Mayeux emphasized that as a diagnostic, it is not ready for prime time. But Trojanowski suggested that it might be very valuable for stratifying subjects in clinical trials. “If you want to do a prevention trial and not wait a long time, take all the people over 60 who have high plasma Aβ levels,” he suggested.—Tom Fagan.
Schupf N, Tang MX, Fukuyama H, Manly J, Andrews H, Mehta P, Ravetch J, Mayeux R. Peripheral Abeta subspecies as risk biomarkers of Alzheimer’s disease. PNAS 2008 September 16;105:14052-14057. Abstract