Measuring plasma Aβ could be a quick, simple, and relatively cheap diagnostic test for conversion from mild cognitive impairment to Alzheimer disease—if the assay was actually informative. Unfortunately, studies on the predictive value of plasma Aβ are equivocal. Is this because the assays being used are inadequate, or is plasma Aβ simply not a good biomarker? A paper in the June Neurobiology of Aging online suggests the latter. Researchers led by Oskar Hansson at Lund University, Sweden, have developed a highly sensitive assay for plasma Aβ and used it to measure the peptide in blood samples from MCI patients who were followed for conversion to AD. “Our hope was that this was a new type of method for plasma Aβ that was more sensitive and that would show some clear differences between different diagnostic groups,” said Kaj Blennow, one of the principal authors on the study, in an interview with ARF. Instead, the data so far support the conclusion that plasma Aβ is a poor predictor of sporadic AD. Familial AD may be another matter, however. In the May 28 Neurology online, researchers led by John Ringman at the University of California, Los Angeles, report that plasma Aβ42 is higher in the plasma of people with pathogenic mutations in either presenilin or amyloid-β precursor protein and that the levels drop before dementia is apparent, echoing previous findings for CSF Aβ42. The implication is that the drop in plasma Aβ could be used to predict symptom onset in the small number of people with inherited forms of AD.

One of the problems with measuring Aβ is that the peptide is so sticky that other proteins and macromolecules in the plasma bind to it and interfere with analysis. Diluting the sample to drive dissociation of these complexes is one answer to this problem. Another is to add detergents to solubilize the peptide. Hansson and colleagues have done both. Their multiplex assay was developed in collaboration with the biotechnology company Innogenetics, Ghent, Belgium, which specializes in assay systems. The assay is highly sensitive and can measure diluted samples (down to 1:20) simultaneously for Aβ40, Aβ42, and N-terminal truncated forms of both. When Hansson and colleagues diluted the plasma samples, the signal and hence the Aβ concentration actually went up as opposed to down, indicating the release of bound Aβ. “But this didn’t help us find a difference between the diagnostic groups,” said Blennow. It is not clear if the assays measured Aβ oligomers or not, but Blennow told ARF that he does not expect that there are many oligomers in the plasma. “If there were, there would also be Aβ deposits in the peripheral tissues, not just the brain,” he said.

Hansson and colleagues used the assay to measure Aβ species in plasma taken from two cohorts of MCI patients. Follow-up conversion rates were available for four to seven years for the first cohort (117 cases), and for two to four years for the second (110 patients). Forty-one percent of the first cohort converted and 14 percent of the second. There was no correlation between the concentration of any Aβ species in the original samples and conversion.

Several prior studies have addressed the value of plasma Aβ as a diagnostic marker. The data is conflicting as to whether Aβ42 is (see, e.g., Mayeux et al., 2003) or is not (see Blasko et al., 2006) predictive for AD, while data from the Rotterdam population-based study suggest that increased Aβ40 but not Aβ42 predicts susceptibility to AD (Van Oijen et al., 2006). More recently, data from the Cardiovascular Health Study, which began in Pittsburgh in the early 1990s, indicate that plasma Aβ species are at best weak predictors of conversion to AD (see Lopez et al., 2008). “Our data fit in very well with previous studies, depending on how you interpret them,” said Blennow. “Though some studies appear to show opposite effects, when you look at the raw data you see there is a very large overlap between AD patients and controls, or MCI patients that don’t progress [to AD] and those that do,” he suggested.

Douglas Galasko, University of California at San Diego, agreed that the jury is still out on the value of plasma Aβ as a biomarker. Galasko, who has studied plasma Aβ but was not involved in the Hansson et al. work, told ARF that nobody has ever shown that there are cross-sectional differences in plasma Aβ among AD, MCI, and control groups. “In the absence of a cross-sectional difference among these groups, it is difficult to envision that plasma Aβ will turn out to be a strong predictor,” he said.

Galasko also questioned the hypotheses behind plasma Aβ as a biomarker for AD. In the case of CSF Aβ, the rationale that drives the well-established association is that there is aggregation or a failure of clearance in the brain that is selective for Aβ42, and so Aβ42 levels in CSF drop. “In the case of plasma, the closest rationale comes from Steve Younkin and colleagues, in which they claim that a proportion of plasma Aβ, in particular Aβ42, originates from the brain,” said Galasko. If there is a substantial contribution from brain Aβ, then patients who show a decrease in CSF Aβ42 should also show a decrease in plasma Aβ42, and by default there should be a correlation between CSF Aβ42 and plasma Aβ42 levels in MCI or in mild AD, suggested Galasko. “Several groups have looked for this, but no one has reported a significant correlation,” said Galasko. Blennow’s group has even compared plasma and CSF Aβ in patients with severe head trauma, which dramatically and rapidly leads to increases in CSF Aβ. In those patients, plasma Aβ did not change at all (see Olsson et al., 2004). “Plasma Aβ may not be providing a relatively short-term signal as to what is going on in the brain,” suggested Galasko.

Despite the apparent shortcomings, plasma Aβ could have some predictive value if the measurements are taken at the right time. “With the advent of PIB we’ve gained the knowledge that many MCI patients already have a substantial amount of amyloid, so one could take the view that measuring plasma Aβ in MCI is already too late,” said Galasko. Plasma Aβ40 or 42 does not correlate with PIB-PET indices of amyloid burden.

The small population of presymptomatic familial AD mutation carriers may give special insight on this question. Ringman and colleagues from UCLA measured plasma Aβ in 21 subjects—12 people with FAD mutations and nine controls. Eight of the FAD patients were presymptomatic in that they had Clinical Dementia Rating (CDR) scores of zero. The other four of the patients had CDR scores of 0.5. The researchers used an ELISA, but not the Innogenetics test, to measure Aβ. They also measured CSF Aβ and tau in a subset of 11 subjects—four of them controls.

Ringman and colleagues found that the CSF Aβ42:Aβ40 ratio was lower in the FAD subjects than in controls, while total CSF tau and phospho-tau (p-tau181) were higher. Low CSF Aβ42 and high tau are associated with sporadic AD (see ARF related news story) and are also predictive of cognitive decline in normal older adults (see ARF related news story). But in these FAD patients, plasma Aβ also seemed to predict decline. The researchers found that plasma Aβ42 and the Aβ42:Aβ40 ratio were higher in the eight FAD patients who were presymptomatic than in normal controls. Interestingly, those eight patients also had significantly higher plasma Aβ42 than the four patients with CDR 0.5 scores, suggesting that a drop in plasma Aβ42 accompanies symptom onset, though the authors caution that further studies on a greater number of patients are needed to determine if this is indeed the case. This discrepancy between presymptomatic and symptomatic patients might explain why other studies have found no link between plasma Aβ and FAD (see, e.g., de Jonghe et al., 1999).

Both Blennow and Galasko agreed that plasma Aβ could be useful for drug development to monitor the efficacy of anti-amyloid treatment. In the meantime, Galasko suggested that one major effort that could be undertaken to finally answer the question of how predictive plasma Aβ is would be to directly compare data across studies. One problem that has held back progress is that most groups use different assays and different patient samples. “What would really help the field is to have some of the groups that are interested exchange samples, standardize assays, and perform a meta-analysis” he said.—Tom Fagan


  1. A very elegant and interesting study.

    View all comments by Takaomi Saido

Make a Comment

To make a comment you must login or register.


News Citations

  1. Diagnosis of AD—Does Spinal Fluid Hold the Key?
  2. Biomarker Roundup: Collecting Clues from MRIs to RNAs

Paper Citations

  1. . 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.
  2. . Conversion from cognitive health to mild cognitive impairment and Alzheimer's disease: prediction by plasma amyloid beta 42, medial temporal lobe atrophy and homocysteine. Neurobiol Aging. 2008 Jan 1;29(1):1-11. 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. . 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.
  5. . Marked increase of beta-amyloid(1-42) and amyloid precursor protein in ventricular cerebrospinal fluid after severe traumatic brain injury. J Neurol. 2004 Jul;251(7):870-6. PubMed.
  6. . Evidence that Abeta42 plasma levels in presenilin-1 mutation carriers do not allow for prediction of their clinical phenotype. Neurobiol Dis. 1999 Aug;6(4):280-7. PubMed.

Further Reading

Primary Papers

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