For research purposes, investigators define preclinical stages of Alzheimer’s disease by the presence of amyloid and tau biomarkers. Now, the longest published study yet to examine whether these stages are valid confirms that advanced preclinical stages carry the highest risk of progression to dementia. The work highlights the crucial role of tau in cognitive decline. Researchers led by Marilyn Albert at Johns Hopkins, Baltimore, followed a middle-aged, cognitively normal cohort for more than a decade. In the April 11 JAMA Neurology, they report that only people with biomarker evidence of both amyloid and tau pathology subsequently lost memory function. The study has implications for how best to select participants for future prevention trials.

“The finding … is consistent with converging evidence that amyloid is necessary but insufficient to provoke substantial cognitive decline,” wrote Elizabeth Mormino at Massachusetts General Hospital and Kathryn Papp at Brigham and Women’s Hospital, both in Boston, in an accompanying editorial. The presence of tau-mediated neurodegeneration seems to be a critical element for the development of dementia, they added. This emerging awareness refines the interpretation of earlier results, which tied brain amyloid to higher risk of cognitive decline at a population level, but included considerable individual variation (see Dec 2014 conference news). 

Amyloid and Tau Together Drive Decline.

People with only Aβ (gold line) or abnormal tau, i.e. SNAP (dotted line), improve on cognitive tests over time just as controls do (blue line). Those with both biomarkers (gray line) worsen. [Courtesy of Soldan et al., © 2016, American Medical Association. All rights reserved.]

Current diagnostic criteria for AD, commissioned by the National Institute on Aging and the Alzheimer’s Association and first released in 2010, define several stages of preclinical AD (see Apr 2011 news). People with no biomarker evidence of Alzheimer’s are deemed stage 0, while those with evidence of amyloid accumulation are stage 1, and those with both amyloid and tau pathology are stage 2. A stage 3 group displays subtle cognitive deficits in addition to amyloid and tau abnormalities. Later, Cliff Jack and David Knopman at the Mayo Clinic in Rochester, Minnesota, added an additional category, suspected non-AD pathology (SNAP), in which evidence of neurodegeneration occurs in the absence of amyloid (see Aug 2013 conference news). 

Several studies have since found that the likelihood of progressing to dementia increases with each successive stage from 0 through 3 (see Knopman et al., 2012Sep 2013 news). In particular, the combination of amyloid and tau pathology spells trouble, with people in stages 2 and 3 declining fastest on cognitive tests (see Wirth et al., 2013Sep 2014 news). 

Most of these studies had only two to four years of follow-up data, however. To test how the NIA/AA staging criteria hold up over longer timeframes, Albert and colleagues used data from the Biomarkers in Older Controls at Risk for Dementia (BIOCARD) cohort. The National Institutes of Health began this study in 1995, enrolling cognitively healthy middle-aged adults with a family history of AD, and following them for 10 years with regular cognitive testing and collection of cerebrospinal fluid. In 2009, after the NIMH dropped the project, researchers at Johns Hopkins picked it up with funding from NIA​ and continued to test the participants annually for cognition. The test battery comprised the Paired Associates immediate recall and Logical Memory delayed recall from the Wechsler Memory Scale-Revised, the Digit-Symbol Substitution from the Wechsler Adult Intelligence Scale-Revised, and the Boston Naming test. Decline on these measures has been reported to predict future cognitive impairment (see Albert et al., 2014). 

For the study, first author Anja Soldan analyzed 222 BIOCARD participants who provided baseline CSF samples. Of these, 102 were classified as stage 0, 46 as stage 1, 28 as stage 2, and 46 as SNAP, based on their CSF levels of Aβ, total tau, and phospho-tau. Over a mean follow-up period of 11 years, only the stage 2 group declined cognitively. The other groups notched slight improvements on cognitive tests, likely due to practice effects (see image above)

While the results largely agree with previous studies that found steep declines at stage 2, the lack of decline in the amyloid-only and SNAP groups was surprising, researchers noted. In other cohorts, a small percentage of people at stage 1 did develop symptoms of impairment within two to four years. The discrepancy might be due to the young age of the BIOCARD cohort, Soldan told Alzforum. Their average age at baseline was 57, suggesting they might have just begun accumulating amyloid. Other studies have found that abnormal amyloid levels can precede symptoms by 15-20 years (see Apr 2013 conference news). 

In keeping with this idea, BIOCARD participants in stage 2 were slightly older than the other groups at baseline, with an average age of 63, and scored lower on baseline cognitive tests. In addition, about half of the stage 2 group carried an ApoE4 allele, compared with one-third of the cohort overall. ApoE4 did not affect how fast people declined. In line with previous studies, BIOCARD showed that ApoE4 lowers the age of onset for cognitive impairment but not its trajectory, the authors suggested.

What about the SNAP group? Studies conflict on this group’s prognosis, with some progressing and others remaining relatively stable over time (see Aug 2013 conference newsSep 2015 conference news). “The SNAP group is still somewhat elusive in terms of its underlying cause,” Knopman told Alzforum. Knopman is an advisor for the BIOCARD study but was not involved in the current work. To parse out what is happening in the brains of people with elevated CSF tau but no amyloid will require more imaging and autopsy data, he added.

For her part, Soldan noted that some studies reporting progression in the SNAP group defined neurodegeneration using imaging biomarkers such as structural MRI, which may capture a later stage of the disease than elevated CSF tau does. Knopman concurred that CSF tau and structural MRI may measure distinct processes. For example, high CSF tau could reflect localized rather than brain-wide degeneration, which may not cause cognitive problems, he suggested. Conversely, in some people high CSF tau may mark a more benign process, or simply noise in the data, he added.

Researchers agree that larger longitudinal studies in different populations will be needed to nail down the relationship between fluid biomarkers and progression. The BIOCARD cohort were mostly well-educated and white. Albert and colleagues are collaborating to pool BIOCARD data with data from four other similar biomarker studies: the Australian Imaging, Biomarker and Lifestyle study (AIBL), the Adult Children Study (ACS) at Washington University in St Louis, the Imaging Substudy of the Baltimore Longitudinal Study on Aging (BLSA), and the Wisconsin Registry for Alzheimer’s Prevention (WRAP). Together, the five groups have formed the Preclinical AD Consortium. The pooled data will enable researchers to determine if biomarker findings hold across distinct populations and a wider age range.

Biomarkers are increasingly being used to screen participants for trials. The BIOCARD findings imply that screening solely for amyloid could include participants who will not worsen over the course of the trial. In younger cohorts, trialists should select people with both abnormal amyloid and tau, Soldan suggested. Tau screening may be less critical in older groups, whose members are closer to symptom onset, she added. The Anti-Amyloid Treatment in Asymptomatic Alzheimer’s Disease (A4) prevention trial enrolls cognitively normal adults 70 years or older based on a positive amyloid scan (see Jan 2013 news). Because a positive bran scan needs a greater accumulation of amyloid than a positive CSF test does, A4 participants may be more likely to progress than amyloid-positive BIOCARD volunteers were, Soldan predicted. Knopman noted that at present, adding tau screening remains impractical in large trials due to expense and more screen failures.

Commentators saw other intriguing implications in the BIOCARD findings. Because the presence of tau was so closely tied to cognitive decline, tau PET imaging might serve as a surrogate outcome measure in preclinical trials, suggested Stephen Salloway at Brown University, Providence, Rhode Island, in an email to Alzforum. Meanwhile, Mormino and Papp found it encouraging that researchers could discern progression in the middle-aged stage 2 BIOCARD cohort. “Clinical trials targeting stage 2 individuals may be able to detect meaningful cognitive decline at a much earlier age than previously expected,” they wrote.—Madolyn Bowman Rogers


  1. This is an interesting manuscript. The data fits with the diagnostic staging of preclinical AD proposed by the NIA-AA work group in 2011. This extends earlier findings that low CSF Aβ associates with eventual decline in amnestic MCI but that decline occurred sooner in MCI subjects with a combination of low Aβ and high tau (Buchhave et al., 2012). Baseline CSF tau and tau PET may be an important predictors of decline, and change in tau PET may be a key outcome

    A weakness of the  paper is the very wide age range and relatively low mean age of the sample. The young age of this cohort may explain why both stage 1 and ApoE4 were not associated with greater rates of decline.  

    To sum up: Alterations in baseline amyloid signify the presence of the AD disease process; elevated tau may be a good proximity marker for cognitive decline; and changes in tau PET with treatment may be a surrogate marker of clinical benefit.measure in AD clinical trials.


    . Cerebrospinal fluid levels of β-amyloid 1-42, but not of tau, are fully changed already 5 to 10 years before the onset of Alzheimer dementia. Arch Gen Psychiatry. 2012 Jan;69(1):98-106. PubMed.

  2. This is an interesting and important paper on the core AD CSF biomarkers in cognitively normal elderly. Biomarkers are increasingly used for inclusion purposes in clinical trials, to enrich for patients or normal elderly with biomarker evidence of AD pathology. Based on the high concordance between CSF Aβ42 and amyloid PET (Blennow et al., 2015), these biomarkers may be used interchangeably to identify amyloid deposition and to enroll trial participants in anti-amyloid trials.

    Longitudinal biomarker studies in the MCI stage of disease suggest that the injury, or neurodegeneration, biomarkers CSF T-tau and P-tau add to Aβ42 by improving the prediction of progression of symptoms during a clinically relevant timeframe (Buchhave, 2012; van Rossum, 2012). For this and other reasons, the updated International Working Group criteria for AD advocate using the combination of low CSF Aβ42 together with either high T-tau or P-tau for the diagnosis of AD (Dubois et al., 2014). 

    In the present study, out of four hypothetical groups of cognitively normal elderly, only the group with both low CSF Aβ42 and high T-tau or P-tau showed cognitive decline during the clinical follow-up period of 11 years (Soldan, 2016). These findings on normal elderly thus corroborate results of previous studies on the MCI stage of the disease. Importantly, rate of progression was not associated with APOE genotype. As pointed out by the authors, these results are essential for trial design, and suggest that enrollment biomarker criteria should include both abnormal CSF Aβ42 and abnormal T-tau/P-tau levels since such individuals have the greatest likelihood of showing cognitive decline during the trial length. This would optimize the possibility of observing a treatment effect of a drug in trials enrolling cognitively normal individuals.

    Indeed, the novel diagnostic criteria for preclinical AD proposed by Dubois and co-workers make a distinction between the entities “preclinical AD,” meaning AD in the preclinical stage based on both Aβ and tau biomarkers being positive, and “asymptomatic at risk for AD,” meaning cognitively normal people with biomarker abnormalities insufficient to reach the definition of AD, e.g., only amyloid biomarkers being positive (Dubois et al., 2016). 


    . Amyloid biomarkers in Alzheimer's disease. Trends Pharmacol Sci. 2015 May;36(5):297-309. Epub 2015 Apr 1 PubMed.

    . Cerebrospinal fluid levels of β-amyloid 1-42, but not of tau, are fully changed already 5 to 10 years before the onset of Alzheimer dementia. Arch Gen Psychiatry. 2012 Jan;69(1):98-106. PubMed.

    . Advancing research diagnostic criteria for Alzheimer's disease: the IWG-2 criteria. Lancet Neurol. 2014 Jun;13(6):614-29. PubMed.

    . Preclinical Alzheimer's disease: Definition, natural history, and diagnostic criteria. Alzheimers Dement. 2016 Mar;12(3):292-323. PubMed.

    . Hypothetical Preclinical Alzheimer Disease Groups and Longitudinal Cognitive Change. JAMA Neurol. 2016 Jun 1;73(6):698-705. PubMed.

    . Injury markers predict time to dementia in subjects with MCI and amyloid pathology. Neurology. 2012 Oct 23;79(17):1809-16. Epub 2012 Sep 26 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Large Studies Agree: Brain Amyloid Accelerates Cognitive Decline
  2. Revised Diagnostic Criteria for Alzheimer’s Are Published
  3. Biomarkers Predict Alzheimer’s, But Shoe Does Not Always Fit
  4. Paper Alert: Preclinical Alzheimer’s Stages Predict Progression
  5. Together, Aβ and Neurodegeneration Spell Cognitive Decline in Three Years
  6. Longitudinal Data Stand Out at AD/PD Imaging Satellite
  7. Suspected Non-Amyloid Pathology (SNAP)—Not an Open or Shut Case
  8. Suspected Non-Alzheimer Pathophysiology: It’s Not Exactly a Snap
  9. Solanezumab Selected for Alzheimer’s A4 Prevention Trial

Paper Citations

  1. . Short-term clinical outcomes for stages of NIA-AA preclinical Alzheimer disease. Neurology. 2012 May 15;78(20):1576-82. PubMed.
  2. . The effect of amyloid β on cognitive decline is modulated by neural integrity in cognitively normal elderly. Alzheimers Dement. 2013 Mar 7; PubMed.
  3. . Cognitive changes preceding clinical symptom onset of mild cognitive impairment and relationship to ApoE genotype. Curr Alzheimer Res. 2014;11(8):773-84. PubMed.

Further Reading

Primary Papers

  1. . Hypothetical Preclinical Alzheimer Disease Groups and Longitudinal Cognitive Change. JAMA Neurol. 2016 Jun 1;73(6):698-705. PubMed.
  2. . Cognitive Decline in Preclinical Stage 2 Alzheimer Disease and Implications for Prevention Trials. JAMA Neurol. 2016 Jun 1;73(6):640-2. PubMed.