back to news With Sudden Progress, Blood Aβ Rivals PET At Detecting Amyloid thumbnail image

With Sudden Progress, Blood Aβ Rivals PET At Detecting Amyloid


A lot can happen in a year. At the 2017 Alzheimer’s Association International Conference in London, Randall Bateman’s team at Washington University in St Louis wowed the crowd with a blood assay for Aβ that predicted brain amyloid with previously unattainable specificity and sensitivity (Jul 2017 conference news). Coming after nearly two decades of trying, without much success, the mass spec-based technique reignited the quest for a blood test for Alzheimer’s pathology that would equal the diagnostic gold standards of cerebrospinal fluid testing and amyloid PET scans, but less invasively and cheaper, respectively. Soon after, a Japanese group debuted a second highly accurate mass spec technique (Feb 2018 news), and by AAIC 2018, held July 20–26 in Chicago, a competitive field was on full display.

  • A blood test for Alzheimer’s seems within reach.
  • Assays on multiple platforms are getting better at calling brain amyloid status.
  • Efforts to standardize are starting.

There was news of a high-sensitivity immunoassay that would be simpler than mass spec and equally good at predicting amyloid positivity. There was a first stab at screening blood with an automated ELISA platform. Scientists presented new data from existing ELISA and mass spec platforms, and an ultrasensitive approach that uses magnets to detect antibody-Aβ complexes on spinning beads. With results rolling in, leaders in the field are already thinking about how to standardize plasma assays, and looking to the pathway they used to bring order to CSF testing (Hansson et al., 2018). 

“Blood tests were the biggest story at AAIC this year,” said Michael Weiner of University of California, San Francisco. “They will become widely available and revolutionize the field in the next few years,” predicted Colin Masters of the University of Melbourne, Australia, who was involved in vetting one of the mass spec assays. “The future vision is that you have the blood test periodically, maybe annually once you are in your 50s, and when your Aβ42/40 ratio changes over two years, then you do something. It could be a trial, a BACE inhibitor, or another safe preventive drug.”

Alzheimer’s disease is defined in part by brain amyloid, which shows up in living people on a PET scan, or can be inferred by a drop in CSF Aβ42 or the Aβ42/40 ratio. But PET is expensive and uses radioactivity, and lumbar punctures are invasive. Neither is quite right for screening and frequent repetitive measures in a clinical trial, or for prescreening in the routine course of healthcare. Indeed, recent early stage trials that enrolled based on biomarkers and no, or only subtle, cognitive signs have struggled with unacceptably high screen failure rates of up to 80 percent.

A blood test would be ideal: With a simple draw, people could have a measure of their likelihood of harboring brain amyloid. While that information could be used to enrich clinical trials, the main use of a blood test would be widespread screening in the general population for brain amyloid, said Philip Scheltens, VU University, Amsterdam. “People deemed negative would be sent on their way, while those in the positive range could be candidates for additional testing, possibly including PET and CSF,” he told Alzforum.

The mass spec assays are an option, though samples will have to be sent to central labs for analysis, and some researchers wondered whether they can be scaled up sufficiently for widespread clinical use. Detection by ultrasensitive antibody-based single-molecule array (Simoa) offers an alternative. The Aβ42 and Aβ40 Simoa assays from Quanterix in Lexington, Massachusetts, have proven to reliably pick out the vanishingly low picogram per ml concentrations of the peptides present in plasma. Using this assay, Swedish researchers previously documented lower plasma Aβ40 and 42 in people with AD dementia than controls; people with mild cognitive impairment had intermediate levels. Alas, where CSF Aβ concentrations robustly distinguished amyloid-positive and -negative people in the early, presymptomatic stage of disease, plasma Aβ did not (Janelidze et al., 2016).

In new data presented at AAIC, Inge Verberk at VU University used a Quanterix Simoa test to measure both plasma Aβ peptides and tau in a different group of 248 elders with subjective cognitive decline and known amyloid status, taken from the Amsterdam Dementia cohort. Consistent with the Swedish results, amyloid-positive people had lower plasma Aβ42 and Aβ42/40 ratios compared with amyloid-negative participants. In Verberk’s study, however, the ratio was able to distinguish amyloid status with moderate sensitivity and specificity, with an area under the curve (AUC) of 0.76. Importantly, lower Aβ42/40 was associated with a nearly twofold greater risk of clinical progression to MCI or AD during follow-up. The results suggest that blood assays have prognostic value early in AD, though there is room for improvement.

Better, Best? Of three plasma Aβ assays, in a side-by-side comparison a new prototype on the Simoa platform reportedly achieved higher accuracy than the commercially available Simoa (Quanterix) or the Euroimmun ELISA. [Courtesy of Teunissen lab.] 

Elisabeth Thijssen, who works with Charlotte Teunissen at VU University, presented a promising new version of the Simoa for Aβ40 and 42. Working in collaboration with ADx Neurosciences, Ghent, the scientists switched up the antibodies that are used to capture and detect Aβ42 and 40, and fiddled with the assay conditions to optimize the test for plasma. To pull peptides from serum, they chose the monoclonal antibodies 21F12 and 2G3, which bind to C-terminal residues of Aβ42 and 40, respectively. For detection of the immobilized peptides, they used the pan-peptide 3D6 antibody, which binds to the first five N-terminal residues of both Aβ42 and 40. This differs from the commercially available Quanterix assay, which uses an N-terminal antibody to capture both peptides, followed by 21F12 or 2G3 to detect Aβ42 and 40, respectively.

Thijssen compared the ratio of plasma Aβ42/40 in samples from 20 amyloid-positive AD patients and 20 amyloid-negative people with subjective memory complaints who were matched for sex and age. Amyloid status was assigned on the basis of CSF testing. Because the readings were more tightly clustered in each group, this new assay produced larger group differences and less overlap between groups than a standard ELISA using the same antibodies, or indeed than the Quanterix Simoa kit. The tighter numbers gave a better diagnostic performance: Thijssen showed 95 percent sensitivity and specificity at distinguishing controls from AD for her prototype, compared with 85 and 86 percent for the Quanterix Simoa kit or ELISA assays, respectively. Of the three, the prototype assay also produced the best correlation between Aβ42/40 in plasma and CSF.

In this first feasibility study, the optimized assay performed better than the Quanterix kit at distinguishing amyloid-positive from -negative, with an AUC of 0.953 vs 0.852 for the kit, Thijssen reported. The performance was also slightly better than the mass spec assay from the Washington University group (Ovod et al., 2017), and comparable to the top-performing test so far, the Shimadzu mass spec procedure.

Going forward, the Amsterdam group will conduct a larger study on about 400 people, including healthy controls and people with MCI and AD who have amyloid PET data available. Those results will indicate whether the plasma assay correlates with PET, and if it can be used as a screening tool, Thijssen and Teunissen said.

In parallel, ADx NeuroSciences will continue to develop and validate the assay format, including scale-up. Both groups will collaborate to commercialize the assay, which generated intense interest at the meeting from researchers involved in many of the large clinical cohorts, Teunissen told Alzforum. The talk also raised the interest of representatives from the Bill Gates Foundation’s new Alzheimer’s initiative (Jul 2018 news), who were spotted chatting with Thijssen and Teunissen after the session.  

Existing plasma assays are getting new attention, as well. Araclon Biotech of Zaragoza, Spain, offered a blood test for Aβ peptides four years ago, but at first found few takers, Araclon’s Ian Sherriff told Alzforum. Their immunoassay tots up free Aβ42 and 40, as well as what they call “total Aβ,” a fraction that includes peptides bound to plasma proteins and red blood cells. The bound fractions are not detected in ELISA or Simoa assays. Araclon scientists originally developed the assay to help along the clinical trial of their anti-amyloid antibody, and then began to offer analysis on a contract service basis. Sherriff told Alzforum that this business has been picking up in the past year.

“It was difficult to convince people we had a decent signal,” Sherriff said. “Now that other groups are coming out with data, that helps a lot. The whole mindset in industry is that blood-based biomarkers are back in,” he told Alzforum.

At AAIC, Shannon Risacher from Liana Apostolova’s lab at Indiana University School of Medicine, Indianapolis, showed how she used the Araclon service to compare blood Aβ to imaging measures of amyloid, tau, and neurodegeneration. She compared Aβ42 and 40 in 47 volunteers who had structural MRI and amyloid PET imaging; 41 also had tau PET with flortaucipir. The group included 20 cognitively normal people, 11 with MCI and 16 with AD. As expected, Aβ42/20 ratios were reduced in patients with AD. The Araclon test distinguished amyloid-positive and-negative in the full sample, with an AUC of 0.8. Among cognitively normal people, the separation was better, with an AUC of 0.865.

Looking at regional amyloid accumulation, Risacher found that a lower plasma Aβ42/40 ratio was linked to increased brain amyloid across the cortex, most strongly in the frontal, parietal, and temporal lobes. The study showed, for the first time, an association of blood Aβ with tau: In this group, total Aβ42/40 negatively correlated with total levels of cortical tau, and more specifically with tau deposition in lateral temporal, inferior parietal, and parietal regions. The ratio’s association with neurodegeneration was much weaker. Risacher detected trends, but no significant association with hippocampal CA1 or CA2-3 volume or subiculum volume. “This shows an association of blood Aβ with specific pathology,” she told Alzforum. “It’s strongest with amyloid, next with tau and weakest with neurodegeneration, which again suggests more sensitivity early in the disease process,” she said.

Risacher would like to repeat her analysis in a larger sample, but also thinks this initial group offers an opportunity to compare different assays head to head. She has sent samples to Henrik Zetterberg, University of Gothenburg, Sweden, for analysis with his lab’s Simoa assay.

Researchers in Koichi Tanaka’s mass spectrometry research lab at Shimadzu Corporation, Kyoto, Japan, continue to refine their method, which combines immunoprecipitation of Aβ peptides with quantitative MALDI-TOF mass spec. As published earlier this year, a composite biomarker based on Aβ42 and 40 blood concentration was 87.5 percent concordant with PiB PET, using a mean cortical SUVR of 1.4 as the cutoff for positivity. At AAIC, Akinori Nakamura in the Tanaka lab presented new data on discordant cases, where PET and blood measures disagreed. He focused on 10 cases where the SUVR was above 1.30 but just shy of the cutoff. All 10 were deemed positive by visual read of their PET scan, and six were classified positive based on the blood biomarkers. Seven of the 10 had follow-up imaging within 12 to 60 months, and by that time four of them crossed the SUVR cutoff. These results indicate that the plasma biomarker is sensitive enough to detect subthreshold amyloid accumulation, Nakamura concluded.

That jibes with data presented by Sebastian Palmqvist, Lund University, Sweden. Palmqvist ran a cross-sectional analysis of changes in a host of biomarkers in the Swedish BioFINDER study, showing that CSF Aβ42 dropped first, followed by the Aβ42/40 ratio. Decreases in plasma Aβ42 and the 42/40 ratio trailed closely behind, at a time when PET signals were ramping up in early amyloid-accumulating regions, but had not yet crossed the threshold of brain-wide amyloid positivity.

Naoki Kaneko of Shimadzu said the Tanaka lab is planning collaborations with sites in the U.S. and Europe to further test their assay. The company is planning to offer commercial analysis services, possibly within this year, he said.

Another entrant into the plasma biomarker arena, MagQu, New Taipei City, Taiwan, markets a unique platform to assay plasma Aβ42, 40, tau, p-tau and α-synuclein. Their system uses a single antibody conjugated to magnetic nanobeads to capture each protein, and a magnetic spin technology to separate and detect antigen-antibody complexes (Apr 2018 conference news). Other available ultrasensitive assays use two antibodies, or one antibody and mass spec for detection. Unlike the sandwich assays or mass spec that detect a decrease in plasma Aβ42 in people with AD dementia, the MagQu assay finds plasma Aβ42 increased or unchanged (Teunissen et al., 2018; Lue et al., 2017). “This is puzzling, but appears very reproducible,” Zetterberg told Alzforum. He speculated that the MagQu assay may be detecting Aβ aggregates and/or Aβ bound to other proteins. Consistent with this idea, Korean researchers recently reported an increase in plasma Aβ42 in people with AD using an oligomer-specific ELISA (Wang et al., 2017). Teunissen told Alzforum her group is planning a side-by-side comparison of that ELISA with the MaqQu and her new Simoa. “Hopefully, that will provide a more conclusive comparison of value of the tests,” she said.

At AAIC, Lih-Fen Lue of Arizona State University showed new data using the method to analyze plasma Aβ42, 40, and tau in 368 healthy people from ages 24–91 from Taiwan, China, Sweden, and the U.S. Plasma Aβ42 and 40 showed a slight decrease with age, but the correlation was weak. She did see a correlation between Aβ42, 40, and t-tau that appeared in middle age and was strongest at ages 60–69. The company has opened a sample testing lab in Arizona to provide the assay in the U.S.

The Global Biomarkers Standardization Consortium (GBCS), led by University of Gothenburg’s Kaj Blennow and convened by Alzheimer’s Association, has led the effort to standardize CSF biomarkers, and in Chicago, the group considered moving quickly to do the same for plasma biomarkers. Michelle Mielke, Mayo Clinic, Rochester, Minnesota, leads the Association’s Biofluid-Based Biomarkers interest group; she updated the GBSC on plasma biomarkers at their preconference, where there was also a panel discussion on plasma Aβ. “There is a lot of interest in plasma Aβ. We need to figure out who is working on what, so we can identify means of collaborating and not recreating the wheel. To my knowledge, there is no consensus yet on what platform should be used for plasma amyloid, but if we want to use it as a screening measure, it does need to be made to be high throughput,” Mielke told Alzforum.

Could one such assay be the Elecsys fully automated immunoassay? Oskar Hansson at Lund University had previously shown that Roche’s automated assay for CSF Aβ42, total-tau, and phospho-tau181 was just as accurate at assigning amyloid status as a visual read of an amyloid PET image by an experienced clinician (Hansson et al, 2018; Apr 2017 conference news). 

To find out how the automated immunoassay would work with blood, Hansson collaborated with Roche to put the system through its paces on 843 plasma samples from the Swedish BioFINDER study. Roche slightly tweaked the assay to optimize it for the lower measuring range of plasma, and incorporated plasma-specific calibrators and controls. Hansson then analyzed biomarkers from CSF and plasma side by side for 513 cognitively unimpaired participants, 265 with MCI, and 65 with AD dementia. Plasma Aβ42/40 ratios were lower in amyloid-positive than amyloid-negative people, regardless of their cognitive state, and the combination of plasma Aβ40 and 42 predicted brain amyloid status with an AUC of 0.80. The accuracies were similar when scientists looked separately at cognitively healthy elderly, or at patients with subjective cognitive decline or MCI.

A composite biomarker comprising plasma Aβ42, Aβ40, and ApoE genotype did a little better, achieving an AUC of 0.85 for brain amyloid status. That was lower than CSF, but higher than the guess of specialists based on a clinical exam and MRI structural imaging of people with subjective cognitive decline (SCD) or MCI (AUC=0.65). Some simple cognitive tests add accuracy to this composite, Hansson said. “Overall, the current sensitivity and specificity of Elecsys plasma tests is not good enough to replace CSF/PET biomarkers, but it was better than the diagnosis by specialist clinicians assessing their patients with SCD or MCI,” Hansson said.

Alas, in primary care, diagnostic accuracy of AD is even lower, Hansson said. He hopes to optimize the Elecsys plasma assay for use in primary care. Even at its current level of sensitivity and specificity, he thinks it could be useful to help primary care practitioners decide whom to refer for specialist neuropsych assessment and CSF or PET testing. However, this needs to be further evaluated in large numbers of people recruited in primary care centers who are less likely to be amyloid-positive, he said.

Tobias Bittner of Roche noted an urgent need for standardization of sample handling for the plasma tests. He said Roche is working on that now, gathering data on the stability of Aβ peptides in plasma and other parameters. “We all need to understand how Aβ behaves in plastic, whether you can transfer from tube to tube or freeze and thaw, the optimal time to centrifuge blood, etc. Preanalytical standardization was a huge problem for CSF, and now we need to look at that for plasma,” he told Alzforum. Bittner said that much of the work is done and promised to share results soon.

Zetterberg and Blennow are coordinating a round-robin study involving all available plasma Aβ methods, in which aliquots of the same samples will be sent around to labs performing the different assays. “We can then compare how well they correlate and what their variations are. We will also examine their potential diagnostic utility by calculating the ratio of amyloid-positive and -negative samples, with amyloid positivity determined by PET,” Zetterberg wrote in an email to Alzforum. They aim to include samples from 20 amyloid-positive and 20 amyloid-negative people. The challenge is to get large enough volumes to provide aliquots for all the labs, he said.—Pat McCaffrey

News Citations

  1. Closing in on a Blood Test for Alzheimer’s? 2 Feb 2018
  2. $30 Million Venture Philanthropy Project for AD Biomarkers 19 Jul 2018
  3. After Plasma Aβ, Now Plasma P-Tau181 Shows Promise 5 Apr 2018
  4. Are CSF Assays Finally Ready for Prime Time?15 Apr 2017

Paper Citations

  1. Hansson O, Mikulskis A, Fagan AM, Teunissen C, Zetterberg H, Vanderstichele H, Molinuevo JL, Shaw LM, Vandijck M, Verbeek MM, Savage M, Mattsson N, Lewczuk P, Batrla R, Rutz S, Dean RA, Blennow K. The impact of preanalytical variables on measuring cerebrospinal fluid biomarkers for Alzheimer’s disease diagnosis: A review.Alzheimers Dement. 2018 Jun 22; PubMed.
  2. Janelidze S, Stomrud E, Palmqvist S, Zetterberg H, van Westen D, Jeromin A, Song L, Hanlon D, Tan Hehir CA, Baker D, Blennow K, Hansson O. Plasma β-amyloid in Alzheimer’s disease and vascular disease.Sci Rep. 2016 May 31;6:26801. PubMed.
  3. Ovod V, Ramsey KN, Mawuenyega KG, Bollinger JG, Hicks T, Schneider T, Sullivan M, Paumier K, Holtzman DM, Morris JC, Benzinger T, Fagan AM, Patterson BW, Bateman RJ. Amyloid β concentrations and stable isotope labeling kinetics of human plasma specific to central nervous system amyloidosis. Alzheimers Dement. 2017 Aug;13(8):841-849. Epub 2017 Jul 19 PubMed.
  4. Teunissen CE, Chiu MJ, Yang CC, Yang SY, Scheltens P, Zetterberg H, Blennow K. Plasma Amyloid-β (Aβ42) Correlates with Cerebrospinal Fluid Aβ42 in Alzheimer’s Disease. J Alzheimers Dis. 2018;62(4):1857-1863. PubMed.
  5. Lue LF, Sabbagh MN, Chiu MJ, Jing N, Snyder NL, Schmitz C, Guerra A, Belden CM, Chen TF, Yang CC, Yang SY, Walker DG, Chen K, Reiman EM. Plasma Levels of Aβ42 and Tau Identified Probable Alzheimer’s Dementia: Findings in Two Cohorts. Front Aging Neurosci. 2017;9:226. Epub 2017 Jul 24 PubMed.
  6. Wang MJ, Yi S, Han JY, Park SY, Jang JW, Chun IK, Kim SE, Lee BS, Kim GJ, Yu JS, Lim K, Kang SM, Park YH, Youn YC, An SS, Kim S.Oligomeric forms of amyloid-β protein in plasma as a potential blood-based biomarker for Alzheimer’s disease.Alzheimers Res Ther. 2017 Dec 15;9(1):98. PubMed.
  7. Hansson O, Seibyl J, Stomrud E, Zetterberg H, Trojanowski JQ, Bittner T, Lifke V, Corradini V, Eichenlaub U, Batrla R, Buck K, Zink K, Rabe C, Blennow K, Shaw LM, Swedish BioFINDER study group, Alzheimer’s Disease Neuroimaging Initiative. CSF biomarkers of Alzheimer’s disease concord with amyloid-β PET and predict clinical progression: A study of fully automated immunoassays in BioFINDER and ADNI cohorts. Alzheimers Dement. 2018 Mar 1; PubMed.