Are cerebrospinal fluid biomarkers for Alzheimer’s disease ready to branch out from research centers into clinical care? Thus far, high variability of CSF measurements, and lingering doubt as to what an abnormal reading means for a particular patient, have restricted this technique to research use in the United States. Now, new evidence suggests that with rigorous quality control procedures, Aβ42 CSF measurements can indeed be reliable and predictive enough to aid diagnosis. In the August 25 JAMA Neurology, researchers led by Oskar Hansson at Skåne University Hospital, Malmö, Sweden, in collaboration with Henrik Zetterberg and Kaj Blennow at the University of Gothenburg, Sweden, reported that low CSF Aβ42 distinguishes cognitively impaired patients who have brain amyloid accumulation from those who do not.

Currently, only PET imaging tracers are approved by the Food and Drug Administration to determine the presence of brain amyloid. Routine use of fluid biomarkers might have advantages over PET imaging, being cheaper, faster, and easier for some clinics to perform. Clinicians would still need to combine the results with clinical histories, cognitive tests, and structural brain imaging to diagnose the disease, Hansson stressed.

“This is one of the first studies to compare imaging and fluid biomarkers in a large set of data,” said Hugo Vanderstichele at ADx Neurosciences, a biotech company based in Ghent, Belgium, that develops diagnostics for neurodegenerative disease. Vanderstichele previously worked at Innogenetics, the maker of the INNOTEST assay used in Hansson’s study. Vanderstichele was not involved in the research. He noted that the findings are encouraging, but that the diagnostic cutoff value for CSF Aβ42 will have to be confirmed by additional studies.

Previous studies reported good correlation between low CSF Aβ and high brain amyloid load (e.g., Jan 2006 news story on Fagan et al., 2006; Grimmer et al., 2009; and Landau et al., 2013). Moreover, low CSF Aβ predicts progression to Alzheimer’s disease (see Jun 2009 news story; Shaw et al., 2009Jan 2012 news story; and Sep 2013 news story). However, these studies did not examine whether CSF Aβ could distinguish between amyloid-positive and -negative people in routine clinical practice.

Amyloid imaging and CSF Aβ 42 agree 92 percent of the time when classifying memory clinic patients as either amyloid-negative (tan rectangle) or amyloid-positive (blue rectangle). [Image copyright © 2014 American Medical Association. All rights reserved.]

To investigate this, first author Sebastian Palmqvist looked at data from 118 patients between 60 and 80 years old who were seen at one of three memory clinics in Sweden over two years. The patients took part in the Swedish BioFINDER (Biomarkers For Identifying Neurodegenerative Disorders Early and Reliably) Study. After initial cognitive testing, about half the patients were diagnosed with subjective cognitive decline, a third with amnestic mild cognitive impairment (MCI), and the remainder with non-amnestic MCI. All patients had lumbar punctures to sample CSF Aβ42, tau, and phosphorylated tau. Unlike the United States and most countries worldwide, in Sweden CSF AD biomarkers are approved for clinical diagnostic use. Participants also underwent PET amyloid scanning with the tracer 18F-flutemetamol.

Scans revealed that half the patients had pathological amyloid accumulation in the brain. Nearly everyone in this group also had low CSF Aβ42. A cutoff of 647 pg/ml or less identified amyloid-positive patients with a sensitivity of 95 percent and a specificity of 90 percent, the authors found. To put it another way, CSF Aβ and amyloid imaging classified 92 percent of the cohort identically (see image above). The authors then used this Aβ42 cutoff in a validation cohort of 38 more clinic patients, where it performed similarly. The cutoff worked equally well to predict brain amyloid in people with either amnestic or non-amnestic MCI or subjective cognitive complaint. Tau and phosphorylated tau correlated only moderately with brain amyloid.

Low CSF Aβ42 associated with amyloid deposition in all brain regions; the link was tight in areas of intense amyloid deposition, such as cingulate cortex and precuneus, and weaker in areas of low deposition, such as medial temporal cortex. This suggests that CSF Aβ42 reflects total brain plaque burden, the authors note. However, there was no correlation between how low CSF Aβ42 was and how high the amyloid burden was. This is likely because, as others including Clifford Jack at the Mayo Clinic in Rochester, Minnesota, have proposed (see Jan 2010 webinar), CSF Aβ42 bottoms out before AD symptoms appear, and thus cannot be used to further stage the disease, whereas brain amyloid continues to accumulate slowly into the early symptomatic phase, the authors suggest.

In ongoing work, Hansson is investigating whether the ratio of CSF Aβ42/Aβ40, or Aβ42/Aβ38, might correlate more closely with brain amyloid than does Aβ42 alone. Vanderstichele agreed that this approach might improve results because it compares pathological Aβ against the total pool of the peptide.

Despite the strong correspondence between brain amyloid and low CSF Aβ, Hansson cautioned that CSF by itself should not be used to diagnose AD, particularly in people whose cognition is normal. Rather, he said, it could be used to rule out Alzheimer’s in people with memory impairment, or to support a clinical diagnosis of AD, similar to the way amyloid imaging is used.

A key feature of this study was the use of stringent quality control standards developed by Blennow and Zetterberg as part of the Alzheimer’s Association’s worldwide Quality Control Program (see Nov 2009 news story). Among other things, the guidelines call for the use of specific polypropylene tubes, handling all samples exactly the same way, using internal standards, and running standards from old and new batches in parallel to ensure they are comparable. With these procedures, variability of standard samples stayed within 7 to 11 percent across the three clinics in this study. “If clinical labs around the world started to use these methods, then standardization would improve immediately,” Zetterberg noted. Overall, this is not yet the case, and variability between labs remains significantly higher, according to the latest data of the quality control program (see Mattsson et al., 2013). However, in the latest, unpublished round of data, variability has begun to drop, Zetterberg told Alzforum.

Zetterberg and Blennow also co-lead the Global Biomarkers Standardization Consortium (see Aug 2012 news story), which works toward assays that perform as robustly worldwide as blood insulin or statin tests. The GBSC has recently submitted two reference measurement procedures for CSF Aβ for approval by the Joint Committee for Traceability in Laboratory Medicine (JCTLM) (see May 2014 news storyKorecka et al., 2014) and is testing a candidate CSF Aβ42 reference material. Once approved and adopted, those advances are expected to further reduce variability, either with current CSF assays or new ones being developed.

One company, Euroimmun, already commercializes fully automated second-generation ELISAs, while other companies, including Roche Diagnostics in Basel, Switzerland, Saladax Biomedical in Bethlehem, Pennsylvania, are developing automated methods for analyzing CSF Aβ42 and tau, as well. Researchers agreed that such automated systems are likely to be more accurate and replace current assays, but noted that any new methods must also be validated against brain imaging before moving to clinical use.—Madolyn Bowman Rogers


No Available Comments

Make a Comment

To make a comment you must login or register.


News Citations

  1. Brain Imaging Speaks Volumes about AD and the Aβ Sink
  2. Subjective Memory Complaint—Not So Subjective Anymore?
  3. Research Brief: More Evidence That CSF Aβ Changes Precede AD
  4. Paper Alert: Preclinical Alzheimer’s Stages Predict Progression
  5. Worldwide Quality Control Set to Tame Biomarker Variation
  6. CSF Markers: Goodbye, Research Use Only; Hello, Clinical
  7. More Accurate Ways to Measure CSF Aβ Debut

Webinar Citations

  1. Together at Last, Top Five Biomarkers Model Stages of AD

Paper Citations

  1. . Inverse relation between in vivo amyloid imaging load and cerebrospinal fluid Abeta42 in humans. Ann Neurol. 2006 Mar;59(3):512-9. PubMed.
  2. . Beta amyloid in Alzheimer's disease: increased deposition in brain is reflected in reduced concentration in cerebrospinal fluid. Biol Psychiatry. 2009 Jun 1;65(11):927-34. PubMed.
  3. . Comparing PET imaging and CSF measurements of Aß. Ann Neurol. 2013 Mar 28; PubMed.
  4. . Cerebrospinal fluid biomarker signature in Alzheimer's disease neuroimaging initiative subjects. Ann Neurol. 2009 Apr;65(4):403-13. PubMed.
  5. . CSF biomarker variability in the Alzheimer's Association quality control program. Alzheimers Dement. 2013 May;9(3):251-61. PubMed.
  6. . Qualification of a surrogate matrix-based absolute quantification method for amyloid-β₄₂ in human cerebrospinal fluid using 2D UPLC-tandem mass spectrometry. J Alzheimers Dis. 2014;41(2):441-51. PubMed.

External Citations

  1. Biomarkers For Identifying Neurodegenerative Disorders Early and Reliably
  2. Joint Committee for Traceability in Laboratory Medicine

Further Reading


  1. . Diagnostic performance of a CSF-biomarker panel in autopsy-confirmed dementia. Neurobiol Aging. 2008 Aug;29(8):1143-59. Epub 2007 Apr 10 PubMed.

Primary Papers

  1. . Accuracy of brain amyloid detection in clinical practice using cerebrospinal fluid β-amyloid 42: a cross-validation study against amyloid positron emission tomography. JAMA Neurol. 2014 Oct;71(10):1282-9. PubMed.