5 April 2012. Researchers have been measuring Aβ for decades, trying to nail down its behavior in Alzheimer's patients or those nearing the disease. Some studies say plasma Aβ rises, some say it falls, and some say it doesn't change at all. Is there any commonality among them? Perhaps, suggests a meta-analysis of 13 plasma Aβ studies, published online March 26 in the Archives of Neurology. "Individually, Aβ42 and Aβ40 did not predict Alzheimer's or dementia very well," said Alain Koyama, of the University of California, San Francisco, who led the study. "But when we looked at the ratio of Aβ42 to Aβ40, we found that a lower value predicted dementia or AD." The report does not mean the field should jump to use the ratio as a biomarker of Alzheimer's disease (AD), scientists agreed. But it may "give some insight to the relationship between the central nervous system and peripheral compartments that is reflective of important biology in the brain," said Ron Demattos of Eli Lilly and Company in Indianapolis, Indiana, who did not participate in this study.
In anticipation of a preventative treatment for Alzheimer's disease, many researchers are looking for ways to screen large populations for at-risk individuals. While cerebral spinal fluid markers have increasingly robust data behind them (see ARF Webinar), “a blood test would be more viable [for large population screens] since it's fairly accessible, less invasive, and cheap," said Koyama.
"It's an admirable study, in that the authors were able to successfully undertake a complicated meta-analysis,” said Anne Fagan, Washington University School of Medicine in St. Louis, Missouri. "Blood-based markers are tricky...and every possible combination of results has been reported," she added, meaning that estimates of plasma Aβ species' levels in relation to dementia and AD are all over the map—elevated, diminished, and unchanged. The paper does not add news to the field's thinking, she said, but it does sum up the variation in research results and points to a lowered Aβ42:Aβ40 ratio as a biomarker that warrants evaluation in a standardized, well-designed study.
Koyama, who completed the research while he was at Harvard School of Public Health, systematically searched three databases—MEDLINE, EMBASE and PsycINFO—for epidemiological studies that measured at least one plasma Aβ species and reported an association with dementia, AD, or cognitive decline. Out of 726 possibly relevant titles captured by that wide net, 13 were included in the meta-analysis. These all involved humans, measured plasma Aβ, were prospective in design, and reported some type of risk ratio. Putting the results together, neither Aβ40 nor Aβ42 consistently forecast cognitive decline. However, the majority of reports that looked at the ratio of Aβ42:Aβ40 in the blood found that participants with ratios in the bottom quantile had a higher risk of AD and dementia than those in the top quantile. Koyama calculated pooled risk ratios for dementia (1.67) and AD (1.60).
This does not make plasma Aβ a strong candidate biomarker for AD, Koyama and outside researchers agreed. When Koyama extracted risk estimates for each study, some estimates were high and others were low, contributing to so much variability that the results should be cautiously interpreted, the authors wrote. "The main limitation of the study was the heterogeneity, which we mostly attribute to studies using different assays," said Koyama. Inconsistent detection methods likely picked up dissimilar Aβ levels among studies, the authors concluded. Widespread failure to screen for baseline mild cognitive impairment may also have contributed to differing results among studies.
Why is plasma Aβ so difficult to pin down? In general, even in the cerebrospinal fluid (CSF), Aβ sticks to itself and to surrounding proteins. But once it crosses into the blood, potential sticking partners abound even more, and the protein gets degraded quickly, Fagan explained. While plasma Aβ is not promising as a biomarker right now, more work is needed to decide its candidacy once and for all, she told Alzforum. "I think the definitive, scientifically most rigorous study has not been performed yet," she said. The ideal analysis would involve a large sample size, standardized collection protocol (controlling for factors such as patient fasting and time of day), effective plasma Aβ assay, and careful evaluation of people's cognitive status using standard criteria.
Lars Lannfelt, Uppsala University, Sweden, agreed that improvements and standardization in assays and sample storage would improve comparability. "We should do [the analysis] in the same way, so we can compare large numbers of samples," he said. He thinks that the plasma Aβ levels probably do drop in people headed for AD, in parallel with those in the CSF, but better assays need to be developed to detect it.
One potential way to improve detection of Aβ in the blood may be to use antibodies that draw it into the periphery and stabilize its presence there. A clinical trial in progress now, sponsored by Eli Lilly and Company and carried out by researchers at WashU, is looking at whether injection of the Aβ antibody solanezumab into the blood of older adults—both with and without AD pathology and dementia—might lead plasma Aβ to correlate better with amyloid imaging markers. A decade ago, Demattos published that a similar antibody injected into transgenic mice caused plasma Aβ42, Aβ40, and the Aβ40:Aβ42 ratio measurements to closely reflect plaque load in the mouse brains (see ARF related news story on Demattos et al., 2002). It may be that the antibody binds Aβ and temporarily extends its half-life in the blood, allowing for a more robust measurement, he said, though he is not sure whether the same will hold true in humans.
Even if plasma Aβ does not pan out as a biomarker, there may be other ways to detect AD using a blood test in the future. For instance, Thomas Kodakek's lab at The Scripps Research Institute in Jupiter, Florida, detected three antibodies that only show up in serum of people with the disease (see ARF related news story).—Gwyneth Dickey Zakaib.
Koyama A, Okereke OI, Yang T, Blacker D, Selkoe DJ, Grodstein F. Plasma Amyloid-β as a predictor of dementia and cognitive decline. Arch Neurol 2012 March 26. Abstract