One marker of Alzheimer’s in the brain is bad news, but two spell real trouble in cognitively normal older adults, according to a paper in the September 15 JAMA Neurology. Scientists led by Reisa Sperling, Harvard Medical School, Boston, found that people with both brain amyloid and neurodegeneration performed worse on neuropsychological tests over the next three years. People who had neither biomarker improved over this time period due to practice, and people with only one biomarker improved a little—evidence of a weaker practice effect.

“This study shows that both neurodegeneration and amyloid deposition independently have an impact on global cognition, and that together they have a synergistic effect,” Gaël Chételat, INSERM-EPHE-University of Caen, France, wrote to Alzforum in an email (see full comment below). The findings also imply that practice effects could be useful for detecting subtle deterioration in people who otherwise appear healthy, wrote the authors. To Sperling, the findings suggest that people with amyloid deposition plus neurodegeneration might decline fast enough for a treatment effect on cognition to be measurable within the time span of a secondary prevention trial, but people with only amyloid deposition might not.

It is known that cognitively healthy people with brain amyloid deposition or evidence of neurodegeneration—be it hippocampal shrinkage, low glucose metabolism in certain areas, or high amounts of phosphorylated tau in the cerebrospinal fluid—are likelier to decline cognitively than people with just one biomarker (see Knopman et al., 2012Desikan et al., 2012Vos et al., 2013). Researchers disagree about whether these markers act independently (see Chetelat, 2013Knopman et al., 2013). Sperling and colleagues wanted to probe this question in cognitively normal people. They also wanted to test whether they could pick up evidence of cognitive decline over two to three years. Secondary prevention trials for Alzheimer’s disease, such as A4 (see Sep 2013 news story) may be too short to use clinical progression as an endpoint, first author Beth Mormino explained.

The researchers enrolled 166 cognitively normal participants from the Harvard Aging Brain Study. Their median age was 74. All scored in the normal range on neuropsychological tests such as the Mini-Mental State Examination (MMSE), Clinical Dementia Rating Scale, and Geriatric Depression Scale. To distinguish those who were positive or negative for amyloid deposition, the researchers used positron emission tomography with Pittsburgh Compound B. To identify neurodegeneration, they measured reduced glucose metabolism with FDG-PET, and hippocampal atrophy by volumetric MRI. Participants with either were deemed positive for neurodegeneration. 

The cohort fell into four groups: 81 participants with neither amyloid nor neurodegeneration; 19 with amyloid only; 28 with both; and 38 for neurodegeneration without amyloid. These groups correspond to the proposed new staging method for preclinical AD, and represent Stages 0, 1, 2, and what some scientists call  “suspected non-AD pathophysiology,” aka SNAP (see Sperling et al., 2011; Jack et al., 2012). Over the following two to four years, everyone completed annual cognitive tests, including the Logical Memory delayed recall, Selective Reminding Test delayed recall, Face-Name Associative Memory Exam, Trail Making Tests A and B, Digit Symbol Substitution Test, and the MMSE. Scores were rolled into a global composite.

Improvement from year to year, or lack thereof, best distinguished the groups. These practice effects arise on repeat testing as a person becomes familiar with the questions, grows more comfortable taking the test, or remembers information from previously sittings. Stage 2 participants did worse as time went on, while stage 1 participants improved somewhat, but less than stage 0 participants.

The study “support[s] the hypothesis that evidence of both fibrillar amyloid deposition and neurodegeneration is associated with an even greater rate of subsequent cognitive decline,” Eric Reiman, Banner Alzheimer's Institute, Phoenix, wrote to Alzforum in an email (see full comment below). He said that larger cohorts should be followed for longer periods to further clarify how well these biomarkers predict deterioration and clinical progression. 

As clinical trials move toward treating asymptomatic Alzheimer’s disease, researchers are looking for ways to detect subtle cognitive changes (see Jun 2014 news story). The current study suggests a way to do that, said Mormino. “People often see practice effects as sources of noise in their sample, but the lack of them might be a sensitive way to pick up cognitive decline in clinically normal samples,” she told Alzforum. “It’s reassuring that we can detect differences in just two years, since prevention trials may last only three,” she added.

The standardized uptake value cutoff Mormino used in the PiB-PET imaging—1.196— is lower than the 1.4 or 1.5 used in other studies (see Mormino et al., 2014). Sylvia Villeneuve, University of California, Berkeley, recently reported that lower SUVRs better capture people who are at the earliest stages of amyloid deposition (see Jul 2014 conference story). “That almost identical ‘early thresholds’ were found in two independent cohorts supports the idea that amyloid can be detected long before individuals reach widely used SUVR cutoffs,” she wrote to Alzforum in an email. Mormino expected that even if she had used a higher threshold, the main result would hold. “The strongest decline occurred in those with the highest PiB values,” Mormino wrote.

Interestingly, among people whose amyloid deposition was below the threshold, those with the highest values were likelier to exhibit smaller practice effects over time and test positive for neurodegeneration. This could mean that even a little Aβ build-up could be biologically meaningful, the authors wrote. David Wolk, University of Pennsylvania, Philadelphia, agreed. “Some of the SNAP group may represent individuals with cerebral amyloid at levels lower than typically constitutes a positive amyloid scan,” he wrote to Alzforum in an email. “This could reflect that these individuals—mostly non-ApoE4 carriers—have a lower threshold of amyloid required for neurodegenerative changes to occur.”

The findings will not change how researchers recruit for the ongoing A4 trial. A4 selects based on Aβ status, not whether or not people have neurodegeneration. A4 will test whether the monoclonal antibody solanezumab delays Alzheimer’s disease. Investigators will assess post hoc whether the presence of neurodegeneration affected the response to treatment, Mormino said. They will also investigate whether certain cognitive domains are more sensitive to changes in practice effects than others. “If we understood where the practice effects are coming from, we could try to exaggerate them to our advantage,” she said.

“It is helpful that we are building up datasets in this preclinical window,” said Nick Fox, University College London. “As we investigate earlier periods of disease, we will need to understand not just the predictive value of these cross-sectional and longitudinal changes, but figure out how they vary and if they can be used as outcome measures for trials,” he said. He noted that a drop in practice effects constitutes one of the earliest cognitive changes he sees in people who carry mutations for familial Alzheimer’s.

The findings “highlight the importance of examining continuous cognitive outcomes in cognitively normal individuals” to detect practice effects, wrote Susan Resnick, National Institute on Aging, Baltimore, in an accompanying editorial.—Gwyneth Dickey Zakaib


  1. This is a very nice study: informative, useful, and timely. It shows that while both neurodegeneration and amyloid deposition independently have impacts on the global cognitive trajectory at follow-up, they have a synergistic effect such that their concomitant presence accelerates cognitive decline. This study emphasizes the view that the biomarkers may be considered independently (rather than sequentially), with "the additive presence of each [biomarker] causing an incremental increase in the risk of (...) progress to AD" (see Chételat, 2013).

    Of course, we have to keep in mind that the findings may differ with longer follow-up because the predictive value of each biomarker, and any combination of biomarkers, is likely highly dependent on the duration of the observation. Moreover, different interpretations may be proposed; this study obviously fuels the debate about what should be considered preclinical AD versus what should not. The answer depends on what are considered the physiopathological processes of the disease, and also on its definition, i.e., what should be called "AD" at this early stage.


    . Alzheimer disease: Aβ-independent processes-rethinking preclinical AD. Nat Rev Neurol. 2013 Mar;9(3):123-4. PubMed.

    View all comments by Gael Chetelat
  2. In this elegant study, Beth Mormino, Reisa Sperling, and their colleagues support the hypothesis that biomarker measurements of fibrillar amyloid deposition in cognitively unimpaired older adults are associated with an indicator of subsequent cognitive decline. They also support the hypothesis that biomarker evidence of both fibrillar amyloid deposition and neurodegeneration is associated with an even greater rate of cognitive decline, and that there may be prognostic value in the previously proposed biomarker stages of preclinical Alzheimer’s disease (AD).

    Larger samples and longer follow-up durations are needed to further clarify the extent to which these preclinical stages provide a prognostic indicator of decline and clinical progression. Findings from preclinical AD trials like the ADCS A4 trial are needed to clarify when different amyloid-modifying agents and other investigational agents need to be given in the earlier preclinical AD stages in order to have the most profound therapeutic effect.

    The Alzheimer’s Prevention Initiative Autosomal Dominant (API ADAD) and Apolipoprotein E4 (API APOE4) trials are evaluating different amyloid-modifying agents in cognitively unimpaired persons who, based on their age and genetic background, are at the highest imminent risk for the clinical onset of AD. Since about a third of the individuals in these trials are not expected to meet PET evidence consistent with moderate to frequent neuritic plaques at the time of enrollment, the API trials will evaluate these agents and test the amyloid hypothesis even earlier in the preclinical course of AD, complementing the important efforts now being undertaken in the ADCS A4 and DIAN trials.

    View all comments by Eric M. Reiman
  3. This is a really nice contribution from a group that is obviously at the forefront of work in the preclinical AD domain. A couple of things are unique about this study, but I should note that even the aspects that are consistent with other prior work serve as important replication of the longitudinal implications of different preclinical subtypes, on which we still have very limited data at this point.

    This study is consistent with the notion that the presence of amyloid plus AD-like neurodegenerative change—in this case hippocampal atrophy and/or a signature of FDG-PET measured hypometabolism—represents a likely later preclinical stage relative to the presence of cerebral amyloid without neurodegenerative change, and thus is more likely to be associated with near-term cognitive change.  This has important implications for clinical trial design, as enrichment in this group may enhance the ability to detect a treatment effect using cognitive outcome measures. 

    Another important finding of this report is that SNAP, neurodegeneration in the absence of amyloid, is not a completely benign state, and that these individuals do display some evidence of longitudinal cognitive change relative to those without amyloid and neurodegeneration. As the authors note, this is likely a mixed group with regard to etiology of the neurodegenerative changes they are displaying.  My suspicion is that this heterogeneity in etiology likely results in heterogeneity in outcomes in this group and that further work will be needed to explore who among the SNAP group are more likely to display future cognitive change.

    One of the most interesting findings of this study was that there was a trend for sub-threshold PiB uptake to correlate with neurodegeneration.  In other words, amongst the amyloid-negative individuals, those closer to the threshold of positivity were more likely to display evidence of either hippocampal atrophy or hypometabolism.  This is an interesting finding, as it may suggest that there is some “signal” to the presence of amyloid even below a relatively liberal cutoff and that some in the SNAP group may represent individuals with cerebral amyloid, but at levels lower than typically constitutes a positive amyloid scan. This could reflect that these individuals—mostly non-ApoE4 carriers—have a lower threshold of amyloid required for neurodegenerative changes to occur. Understanding this relationship may have important implications for disease pathophysiology.

    View all comments by David Wolk
  4. This study by Mormino et al. addresses important questions using an elegant methodology. In order to use cognition as an endpoint in clinical trials enrolling cognitively normal older adults, it is important to assess the relationship between the AD biomarkers and cognitive changes and, more importantly, to assess if sufficient cognitive change can be detected during a short follow-up.

    The threshold the authors used to assess amyloid positivity was derived from a Gaussian Mixture Model approach and was set at a standard uptake value ratio of 1.196, which is lower than the widely used SUVR thresholds of 1.4 -1.5 and supports the findings we presented at the last AAIC conference (see Jul 2014 news story). The fact that almost identical “early thresholds” were found in two independent cohorts gives strong support to the idea that amyloid can be detected with PIB-PET imaging long before individuals reach the widely used SUVR cutoffs of 1.4-1.5, and that some subjects classified as PiB-negative in previous studies might in fact have been early amyloid accumulators.

    Looking subthreshold for amyloid positivity is an interesting and novel approach of this paper that will need further investigation. While fixed thresholds are often used in research, it is possible that thresholds vary from one person to another depending on other risks and protective factors. Our previous work suggests, for instance, that less amyloid is needed to detect an association between amyloid and cortical thickness in subjects presenting a vascular burden (see Villeneuve et al., 2014).

    Finally, we need to remember that PiB is a tracer that binds to amyloid, but it is not amyloid per se. Some subjects might therefore already have substantial levels of amyloid before reaching positive PiB-thresholds.


    . Vascular risk and Aβ interact to reduce cortical thickness in AD vulnerable brain regions. Neurology. 2014 Jul 1;83(1):40-7. Epub 2014 Jun 6 PubMed.

    View all comments by Sylvia Villeneuve

Make a Comment

To make a comment you must login or register.


News Citations

  1. NIH Funds Prevention Trials and Translational Studies
  2. Test Battery Picks Up Cognitive Decline in Normal Populations
  3. References and Thresholds—Amyloid Imaging Protocols Debated at AAIC
  4. 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. . Amyloid-β-Associated Clinical Decline Occurs Only in the Presence of Elevated P-tau. Arch Neurol. 2012 Apr 23; PubMed.
  3. . Preclinical Alzheimer's disease and its outcome: a longitudinal cohort study. Lancet Neurol. 2013 Oct;12(10):957-65. PubMed.
  4. . Alzheimer disease: Aβ-independent processes-rethinking preclinical AD. Nat Rev Neurol. 2013 Mar;9(3):123-4. PubMed.
  5. . Selective Worsening of Brain Injury Biomarker Abnormalities in Cognitively Normal Elderly Persons With β-Amyloidosis. JAMA Neurol. 2013 Aug 1;70(8):1030-8. PubMed.
  6. . Toward defining the preclinical stages of Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement. 2011 May;7(3):280-92. Epub 2011 Apr 21 PubMed.
  7. . An operational approach to National Institute on Aging-Alzheimer's Association criteria for preclinical Alzheimer disease. Ann Neurol. 2012 Jun;71(6):765-75. PubMed.
  8. . Amyloid and APOE ε4 interact to influence short-term decline in preclinical Alzheimer disease. Neurology. 2014 May 20;82(20):1760-7. Epub 2014 Apr 18 PubMed.

Further Reading

Primary Papers

  1. . Synergistic effect of β-amyloid and neurodegeneration on cognitive decline in clinically normal individuals. JAMA Neurol. 2014 Nov;71(11):1379-85. PubMed.
  2. . Preclinical biomarkers in Alzheimer disease: a sum greater than the parts. JAMA Neurol. 2014 Nov;71(11):1357-8. PubMed.