For anyone in an Alzheimer’s disease study who is nervous about a spinal tap, the idea of measuring dementia risk with a simple blood test would be a dream come true. This dream has long seemed out of reach. Studies of plasma markers have delivered maddeningly inconsistent results. A paper in the January 19 Journal of the American Medical Association now suggests that a blood test for amyloid-β (Aβ) deserves a closer look. Researchers led by Kristine Yaffe at the University of California in San Francisco took baseline plasma samples from nearly 1,000 elderly normal volunteers and then followed the participants for nine years, regularly measuring their cognitive performance. Unlike similar previous studies, Yaffe and colleagues looked at cognitive decline rather than conversion to AD, perhaps making the results more sensitive. They found that a low ratio of plasma Aβ42 to Aβ40 at baseline correlated with a greater drop in mental abilities over the length of the study. Intriguingly, this association was strongest in participants with low levels of education and literacy, and much weaker in people with more education. This fits with the long-standing view that a person’s cognitive reserve can protect against mental decline. If the result proves robust in future studies, it may re-ignite interest in using blood tests to monitor AD risk.

“The most important message from this and similar studies is that differences in proteins and peptides can be found in peripheral fluids years before clinical onset of dementia. This is good news,” wrote Monique Breteler of University Medical Center Rotterdam, The Netherlands, in an accompanying editorial in JAMA. Breteler points out, however, that to measure up as a biomarker, a protein must be shown to carry a quantifiable disease risk, and that plasma Aβ has a long way to go to meet this standard.

So far in the AD field, cerebrospinal fluid (CSF) biomarkers have dominated the discussion, with low levels of CSF Aβ42 showing a reliable correlation with AD risk (e.g., see ARF related news story). Amyloid in blood, on the other hand, has yielded a more confusing picture. Some studies have found a relationship between low levels of plasma Aβ and development of Alzheimer’s (see ARF related news story on Graff-Radford et al., 2007; Pesaresi et al., 2006; van Oijen et al., 2006; Lewczuk et al., 2010), including one from last year that followed more than 1,000 elderly people in France over four years (see ARF related news story on Lambert et al., 2009). Other work, meanwhile, has shown the opposite pattern, with high levels of plasma Aβ dovetailing with disease risk (see ARF related news story on Schupf et al., 2008; Mayeux et al., 2003). One possible explanation that has been put forth for this discrepancy is that plasma Aβ might be high initially in people at risk for AD, then drop as people approach Alzheimer’s (see Cosentino et al., 2010). This hypothesis does not explain negative results, however: several studies have failed to find any strong association between plasma Aβ and cognition (see ARF related news story on Hansson et al., 2008; Lopez et al., 2008; and Fukumoto et al., 2003).

Part of the problem may lie in the way Aβ is handled, suggested several researchers. “Aβ is difficult to measure,” said Henrik Zetterberg at Sahlgrenska University Hospital in Molndal, Sweden, explaining that the protein not only self-aggregates, but also binds to many other proteins in plasma. “The exact way the assay is performed will influence the results.” To try to minimize this problem, Yaffe and colleagues chose to use the relatively new INNO-BIA assay developed by Innogenetics in Ghent, Belgium, to measure Aβ. This assay was used in the recent French study and is also employed by the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Innogenetics' method is more sensitive than previous techniques, Yaffe said, and appears to deliver consistent results.

The authors also wanted to conduct a large prospective study with extensive follow-up. To that end, Yaffe and colleagues recruited almost 1,000 healthy elderly volunteers living in Tennessee and Pennsylvania; about half were white and half were African-American. The researchers collected blood samples one year into the study, and sent them to the laboratory of Steve and Linda Younkin at the Mayo Clinic in Jacksonville, Florida, for analysis. The results provided baseline values for plasma Aβ that were then divided into tertiles. To monitor cognitive ability, participants took the Modified Mini-Mental State Examination approximately every two years. A low initial ratio of Aβ42/40 was associated with greater cognitive decline over nine years. People in the lowest tertile on average lost more than six points on their exam score, three points more than those in the highest tertile. Adjusting the data for various confounding factors such as age, race, diabetes, smoking, and ApoE genotype made no difference in this result. However, the association between Aβ levels and test scores was weaker in people who had at least a high school diploma and higher than sixth-grade literacy, and stronger in those with less education.

This is one of the most interesting findings from the study, Yaffe said, because it supports the hypothesis of cognitive reserve. Roughly put, the idea is that using your brain more can strengthen it and make it more resistant to neurodegenerative disease. “I think it’s exciting because it implies that even if you have a biomarker [for AD risk], you could do something about it by getting more cognitive stimulation,” Yaffe said. Numerous clinical trials are testing whether mental and physical stimulation can delay the development or progression of cognitive problems or AD (e.g., this cognitive training and exercise trial, cognitive training trial, and computer-based training trial). However, a recent review of cognitive interventions, authored by Mike Martin of the Universität Zurich, Switzerland, and colleagues, found no evidence that trials conducted to date have been effective in improving memory. Martin and colleagues note that it is still possible that longer, more intense, or different interventions might be beneficial (see Martin et al., 2011).

A number of epidemiological studies have shown an association between cognitive reserve and lower dementia risk, but so far only a few have tied this interaction to biomarkers. Two recent papers found that in highly educated people, a heavy brain amyloid load as revealed by positron emission tomography with Pittsburgh Compound B (PIB) was less predictive of mental decline (see Kemppainen et al., 2008; Rentz et al., 2010). Dorene Rentz of Massachusetts General Hospital, Boston, presented new data on this issue at the 2010 Human Amyloid Imaging meeting (stay tuned for upcoming news story), and is enthusiastic about the work by Yaffe and colleagues.

“I think the take-home message is that cognitive reserve does provide a protective performance effect in the earliest stages of the disease and may mask evidence of pathology,” Rentz wrote in an e-mail to ARF, adding, “The strength and value of this study is its nine-year longitudinal follow-up and the large number of subjects who participated.”

The other implication of their work, Yaffe said, is that a blood test for β amyloid may have potential. This idea remains controversial, however. Zetterberg said it is not clear which physical process plasma Aβ levels reflect, since plasma Aβ has not shown a strong correlation with Aβ in the CSF (see Hansson et al., 2008). “It would be great to see how plasma Aβ levels correlate with PIB retention in the brain,” Zetterberg suggested. Yaffe points out that few studies have measured Aβ in both CSF and blood, and more comparisons should be done. This will soon happen, according to Les Shaw at the University of Pennsylvania in Philadelphia, and co-director of the ADNI biomarker core, as ADNI has collected these data and they will be analyzed this year using the INNO-BIA assay. Shaw also emphasized the movement in the AD field toward standardizing study design, including all methods used and performance, which will make comparisons between studies more meaningful.

Researchers agree that it remains to be seen if a blood test for AD will ever be viable. Yaffe noted that their study looked at averaged results for large groups of people. “We don’t know how this translates into an individual’s risk,” she said. Zetterberg speculated that even if the plasma Aβ results are robust and reflect a process happening in the brain, a blood test would probably not be useful as a diagnostic for AD. Instead, such a test would more likely indicate heightened risk, analogous to measuring cholesterol for heart disease.—Madolyn Bowman Rogers


  1. This paper is very interesting, and in keeping with our recent findings using amyloid imaging (see Rentz et al., 2010). I think the take-home message is that cognitive reserve does provide a protective performance effect in the earliest stages of the disease and may mask evidence of pathology. This means that people with higher reserve tend to maintain normal cognitive performance for longer. Unfortunately, if you were attempting to find people with early memory deficits for treatment trials, you would overlook these individuals, as they would continue to perform above cut-offs, despite the disease process. The failure to account for cognitive reserve may be one factor in the negative findings in some studies that could not find an association between Aβ deposition and performance. The strength and value of this study by Yaffe et al. is its nine-year longitudinal follow-up and the large number of participating subjects.


    . Cognition, reserve, and amyloid deposition in normal aging. Ann Neurol. 2010 Mar;67(3):353-64. PubMed.

    View all comments by Dorene Rentz
  2. In this paper, Yaffe et al. describe the results of a prospective study in which 997 persons underwent baseline blood draws and were followed for nine years with serial cognitive assessments using the 3MS. This is an important area of study, as Aβ is thought to play a critical role in driving the AD process in the central nervous system. As the authors point out, however, the relationship between plasma Aβ levels and AD risk is not clear, with prior studies yielding contradictory results; some studies indicate that high and others indicate that low Aβ42 levels put one at risk for cognitive decline and AD. The current study addresses this issue using the largest cohort and the longest follow-up I am aware of. Also of interest is the ethnicity (about 50 percent African-American) and wide range of educational levels represented in the study population.

    Dividing the population into tertiles based on the plasma Aβ42/40 ratios, the authors showed that persons (mean age of 74) with the lowest ratios had the most cognitive decline in the subsequent nine years as assessed with the 3MS. A limitation was that the tertiles differed in such a way that persons in the lowest Aβ42/40 tertile were more likely black, diabetic, had lower literacy, and to have an ApoE4 allele—all things that also contribute to dementia and AD risk. When these variables were controlled for statistically, the effect was diminished but nonetheless remained. The correlation between low plasma Aβ42/40 and AD risk might be interpreted as a shift in equilibrium towards deposition of Aβ in the brain, as has been postulated to explain the decrease in Aβ42 levels in the CSF. This is consistent with some prior studies, including one by our group, looking at persons at risk for familial Alzheimer’s disease (FAD). In our study (involving the smallest number of subjects with the shortest duration of follow-up!), we found that asymptomatic FAD mutation carriers had high levels of plasma Aβ42, while mildly symptomatic mutation carriers (Clinical Dementia Rating Scale score of 0.5) had lower levels (1). This suggests that whether high or low levels of plasma Aβ42 are associated with subsequent cognitive decline may depend on exactly when during the risk period the levels are measured. What continues to be lacking, and is not addressed by the current article, is a mechanistic explanation linking plasma and CSF Aβ42 levels, as they do not typically correlate well.

    Yaffe et al. also demonstrated that “cognitive reserve,” as indexed by literacy and level of education, modified the relationship such that the most cognitive decline occurred in the low reserve, lowest Aβ tertile group. This may be interpreted as indicating that persons with higher reserve are less prone to the toxic effects of Aβ42, though this is far from conclusive.

    As in previous studies, levels of plasma Aβ42 and Aβ40 were highly variable, and there was a great degree of overlap among the tertiles. Also, the size of the effect of the Aβ42/40 ratio on the outcome variable was mild; a three-point difference (on a 100-point scale) in some analyses over a nine-year period. Therefore, though these results are interesting, they bring into question the utility of plasma amyloid measures as clinically relevant predictors of cognitive decline.


    . Biochemical markers in persons with preclinical familial Alzheimer disease. Neurology. 2008 Jul 8;71(2):85-92. PubMed.

    View all comments by John Ringman
  3. The Yaffe paper is indeed very interesting. Finding a diagnostic and/or prognostic plasma biomarker is the real Holy Grail. It's still unclear, however, what alterations in plasma Aβ levels/ratios reflect. Dr. Zetterberg mentioned in the accompanying article that it would be great if we knew whether there was a relationship between plasma Aβ levels and PIB retention in the brain. Our own work in this area (Fagan et al., 2009; Figure 3) has demonstrated no relationship between plasma Aβ (Aβ1-40, Aβx-40, Aβ1-42, Aβx-42) levels and mean cortical PIB binding potential in a large cohort (n = 189) of cognitively normal individuals despite a strong relationship between PIB binding and CSF Aβ42 in this same cohort (as we and now several other groups have shown). Thus, the plasma Aβ story remains elusive...and warranting ongoing study.


    . Cerebrospinal fluid tau and ptau(181) increase with cortical amyloid deposition in cognitively normal individuals: implications for future clinical trials of Alzheimer's disease. EMBO Mol Med. 2009 Nov;1(8-9):371-80. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Triple Confirmation: AD Footprint in CSF of Cognitively Normal People
  2. The Value of Biomarkers—Diagnosis and Genetic Screens
  3. Stayin’ Alive—Huge Study Quickens Quest for Plasma AD Biomarker
  4. Hot Plasma—Blood Aβ a Risk Marker for AD?
  5. Plasma Aβ Testing Beset by Questions of Assays, Biology, Timing

Paper Citations

  1. . Association of low plasma Abeta42/Abeta40 ratios with increased imminent risk for mild cognitive impairment and Alzheimer disease. Arch Neurol. 2007 Mar;64(3):354-62. PubMed.
  2. . Plasma levels of beta-amyloid (1-42) in Alzheimer's disease and mild cognitive impairment. Neurobiol Aging. 2006 Jun;27(6):904-5. PubMed.
  3. . Plasma Abeta(1-40) and Abeta(1-42) and the risk of dementia: a prospective case-cohort study. Lancet Neurol. 2006 Aug;5(8):655-60. PubMed.
  4. . Amyloid beta peptides in plasma in early diagnosis of Alzheimer's disease: A multicenter study with multiplexing. Exp Neurol. 2010 Jun;223(2):366-70. PubMed.
  5. . Association of plasma amyloid beta with risk of dementia: the prospective Three-City Study. Neurology. 2009 Sep 15;73(11):847-53. PubMed.
  6. . Peripheral Abeta subspecies as risk biomarkers of Alzheimer's disease. Proc Natl Acad Sci U S A. 2008 Sep 16;105(37):14052-7. Epub 2008 Sep 8 PubMed.
  7. . Plasma A[beta]40 and A[beta]42 and Alzheimer's disease: relation to age, mortality, and risk. Neurology. 2003 Nov 11;61(9):1185-90. PubMed.
  8. . Plasma ß-amyloid and cognitive decline. Arch Neurol. 2010 Dec;67(12):1485-90. Epub 2010 Aug 9 PubMed.
  9. . Evaluation of plasma Abeta(40) and Abeta(42) as predictors of conversion to Alzheimer's disease in patients with mild cognitive impairment. Neurobiol Aging. 2010 Mar;31(3):357-67. Epub 2008 May 19 PubMed.
  10. . Plasma amyloid levels and the risk of AD in normal subjects in the Cardiovascular Health Study. Neurology. 2008 May 6;70(19):1664-71. PubMed.
  11. . Age but not diagnosis is the main predictor of plasma amyloid beta-protein levels. Arch Neurol. 2003 Jul;60(7):958-64. PubMed.
  12. . Cognition-based interventions for healthy older people and people with mild cognitive impairment. Cochrane Database Syst Rev. 2011;(1):CD006220. PubMed.
  13. . Cognitive reserve hypothesis: Pittsburgh Compound B and fluorodeoxyglucose positron emission tomography in relation to education in mild Alzheimer's disease. Ann Neurol. 2008 Jan;63(1):112-8. PubMed.
  14. . Cognition, reserve, and amyloid deposition in normal aging. Ann Neurol. 2010 Mar;67(3):353-64. PubMed.

External Citations

  1. INNO-BIA assay
  2. Innogenetics
  3. Alzheimer’s Disease Neuroimaging Initiative
  4. cognitive training and exercise trial
  5. cognitive training trial
  6. computer-based training trial

Further Reading

Primary Papers

  1. . Mapping out biomarkers for Alzheimer disease. JAMA. 2011 Jan 19;305(3):304-5. PubMed.
  2. . Association of plasma beta-amyloid level and cognitive reserve with subsequent cognitive decline. JAMA. 2011 Jan 19;305(3):261-6. PubMed.