With the advent of PET amyloid imaging, researchers discovered that as many as a third of cognitively healthy older adults go about their business with brains loaded with amyloid. What does this mean for their cognitive future? Are all these people destined to develop Alzheimer’s disease? Initial studies suggested that brain amyloid greatly ramps up the risk for cognitive decline and a clinical diagnosis of dementia, but it was clear that settling the question would take more years of observational study in larger numbers of people. At the 7th Clinical Trials on Alzheimer’s Disease (CTAD) conference in Philadelphia November 20 to 22, speakers expanded the body of evidence, presenting data from large prospective studies.

All the researchers reported that in their samples of otherwise healthy older adults, people with high brain amyloid levels declined cognitively over periods of two to three years, while those without brain amyloid remain stable. To be sure, the rate of cognitive decline varied from person to person, and some people with brain amyloid did stay sharp over these short time frames. It remains unclear whether these people are simply early in the trajectory of disease, or protective factors operate in their brain to help preserve cognition. Speakers discussed some factors that accelerate or slow disease progression, including ApoE genotype and hypertension. Tantalizingly for trial designers, deeper analysis of the data pouring in from prospective studies of preclinical and prodromal stages suggest that people’s decline on biomarkers and clinical measures during this phase happens in a linear fashion, simplifying the necessary statistics for both disease modeling and trial design.

Regardless of ApoE genotype, older AIBL participants without brain amyloid stayed cognitively stable over 4.5 years. Among those with brain amyloid, ApoE4 carriers declined much faster. [Image courtesy of Paul Maruff.]

Several studies have found that brain amyloid associates with lower baseline cognitive scores and faster decline (see Snitz et al., 2013Mar 2014 news story), particularly when it occurs along with a marker of neurodegeneration (see Sep 2014 news story). However, these studies tended to be small, enrolling 200 people or fewer.

To see if the link would hold up in a larger sample, Ming Lu of Avid Radiopharmaceuticals, a subsidiary of Eli Lilly, analyzed data from 1,017 participants in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) who had undergone a florbetapir PET scan. Lilly manufactures florbetapir. Almost half of the participants had been diagnosed with mild cognitive impairment (MCI) at baseline, and the remainder were either cognitively normal or diagnosed with Alzheimer’s disease. At CTAD, Lu reported that over three years, participants whose brains were free of amyloid stayed stable on cognitive tests, including the ADAS-Cog11, MMSE, and CDR-SB. In contrast, people with positive amyloid scans declined, with MMSE scores diverging significantly from amyloid-negatives at six months, and ADAS-Cog11 scores at 24 months. Moreover, for people with MCI, those with brain amyloid had almost five times the odds of progressing to AD over the course of the study than those without. However, Lu noted that brain amyloid explained only some of the individual variation in decline.

What determines how fast a given person declines? Lilly’s Paula Trzepacz combed through 24-month follow-up data from about 1,200 ADNI participants to hunt for clues. She used statistical analyses to look for subpopulations with different rates of decline, and identified four separate groups. The 470 people without brain amyloid fell into two categories of unequal numbers. About 90 percent remained cognitively stable, and they were equal parts cognitively normal or diagnosed with MCI. The remaining 10 percent of the amyloid-negatives surprised Trzepacz. Their cognitive scores dropped sharply over the study period. Most had been clinically diagnosed with MCI or AD; however, given the absence of amyloid it is likely that this was incorrect, and they suffered from a non-amyloid form of neurodegeneration, Trzepacz suggested. Other studies have identified people with “suspected non-amyloid pathology” (SNAP) as well (see Aug 2013 conference storyNov 2013 news storyNov 2014 news story). This potential SNAP group also had somewhat different symptoms from typical AD, having more difficulty with daily activities than memory, Trzepacz noted.

Trzepacz showed that the 722 people with brain high amyloid fell into two groups. One-quarter of them declined significantly during the study; most of these people had been diagnosed clinically with AD, some with MCI. The remaining three-quarters declined only slightly on the cognitive measures. They included mostly people who had been clinically diagnosed with MCI, and close inspection showed that they had higher cognitive scores at baseline. This cohort may be earlier in the trajectory of Alzheimer’s disease, Trzepacz suggested. The factor that best predicted rapid cognitive decline among amyloid-positive people was a low cognitive score at baseline, pointing perhaps to cognitive reserve as protection against the long-term cognitive effects of brain amyloid. Overall, the presence of amyloid does affect the trajectory of cognitive decline, and this occurs independently of the person’s diagnosis, Trzepacz concluded.

Paul Maruff of Cogstate in Melbourne, Australia, showed data from the Australian Imaging, Biomarker, and Lifestyle Flagship Study of Ageing (AIBL) that supported the hypotheses that factors in addition to amyloid level influence the rate of cognitive decline in early AD. AIBL is one of several prospective imaging and cognition studies that are generating empirical data to test and flesh out the hypothetical Alzheimer’s biomarker staging scheme. AIBL data show that once a person crosses the threshold of amyloid positivity, additional amyloid accumulates at a linear rate for about the next 20 years until Alzheimer’s symptoms appear. Importantly, this finding allows researchers to use linear models to predict progression in prodromal stages—the time period on which preventions trials increasingly are going to focus. “Think of the otherwise sigmoid Jack staging curves without the little bit at the beginning and the little bit at the end,” Maruff quipped.

Maruff analyzed AIBL data from some 200 healthy controls and 50 people with MCI who were being followed over three years. People without brain amyloid remained cognitively stable during this time, regardless of whether they were diagnosed as healthy or with MCI, he reported. On the other hand, people with amyloid all declined at about the same average rate, regardless of diagnosis (see Lim et al., 2014Lim et al., 2014). 

How do the rates of amyloid accumulation and memory loss compare to each other, and to other biomarker change? To relate these very different measures, Maruff calculated the standard deviations of each marker based on the population averages at baseline. He found that in preclinical AD (cognitively normal controls with brain amyloid), amyloid and cognition both worsened by about half a standard deviation over three years. Hippocampal volume shrank only by about one-tenth of a standard deviation. By contrast, in prodromal AD (mild cognitive symptoms and brain amyloid), amyloid, cognition, and hippocampal volume all worsened at the same rate, again half a standard deviation over three years. The acceleration of hippocampal atrophy thus represents a big biomarker difference between the two conditions, Maruff said. He noted that because all three biomarkers change at equivalent rates in prodromal AD, they might all be useful as progression markers in clinical trials.

On the other hand, subjective memory assessments made by patients or caregivers change more slowly, and therefore may be a less-sensitive marker, Maruff added (see Hollands et al., 2015). This finding appears to pour some cold water on recent hopes that subjective memory concerns could be formalized into a useful predictor early on in the symptomatic phase (see Sep 2014 news story). Maruff noted inconsistencies between the AIBL finding that subjective memory may be insensitive to early amyloid-related changes and the findings in other cohorts, for example the Harvard Aging Brain Study, that subjective memory is a sensitive indicator of abnormal amyloid. These, he said, may result from the different ways in which the samples were recruited. Healthy older adults in the AIBL study were recruited from the community and hence may have been less aware of subtle changes in their memory than volunteers recruited from hospital outpatient clinics.

Maruff also clarified that although the average rate of decline is the same in these groups, quite a bit of individual variation occurs. One factor influencing this is ApoE genotype. ApoE4 carriers with amyloid decline faster than non-carriers with amyloid, Maruff said.

Other speakers dug more deeply into the role of ApoE and other modifying factors. For example, Karen Rodrigue of the University of Texas at Dallas had previously reported that ApoE4 carriers with hypertension accumulated amyloid faster than carriers with normal blood pressure, particularly if the hypertension was uncontrolled (see May 2013 news story). That study involved only 118 people, raising the question of whether the findings would hold up in a larger sample.

At CTAD, Rodrigue described her analysis of 1,013 ADNI participants with normal cognition, MCI, or AD. Forty-four percent had at least one copy of ApoE4, and 65 percent had high blood pressure. In agreement with earlier findings, Rodrigue found that people with both risk factors had significantly more amyloid than those who had one or neither. Among the ApoE4 carriers, hypertension nearly doubled the risk of having a positive amyloid scan. In this analysis, a person’s clinical diagnosis did affect the relationship. Specifically, among cognitively normal people, having both risk factors worsened amyloid only if they were over 70. In the MCI group, the interaction between ApoE, hypertension, and amyloid load did not reach statistical significance. In ADNI, the MCI cohort was defined by clinical diagnosis, and biomarker analysis later revealed that many of them did not have brain amyloid. People with AD had the strongest interaction between these factors.

Overall, the data presented at CTAD strengthened the case that amyloid hastens cognitive decline, but also underscored that pure clinical diagnoses poorly capture the underlying pathology. In many current clinical trials, researchers screen potential participants for amyloid pathology or an Aβ/tau CSF signature to ensure that their symptoms are most likely due to Alzheimer’s (see Part 4 of this series).—Madolyn Bowman Rogers.


  1. The SD (or z-score) approach mentioned by Paul Maruff and applied in the AIBL data is indeed useful to compare change on different diagnostic modalities. We have recently applied this approach in ADNI-1 data (Bertens et al., 2014). In prodromal AD we found that hippocampal volume, FDG-PET hypometabolism, and ADAS-Cog score declined simultaneously and the rate of decline further accelerated in subjects with AD-type dementia. Unlike the AIBL study, we did not find change in the amyloid measure, for which we used CSF Aβ1-42, in either preclinical, prodromal, or AD-type dementia. This may suggest that the dynamic of change in CSF Aβ1-42 differs from that of amyloid PET. 


    . Temporal evolution of biomarkers and cognitive markers in the asymptomatic, MCI, and dementia stage of Alzheimer's disease. Alzheimers Dement. 2014 Aug 20; PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Amyvid Follows in PiB’s Footsteps
  2. Together, Aβ and Neurodegeneration Spell Cognitive Decline in Three Years
  3. Suspected Non-Amyloid Pathology (SNAP)—Not an Open or Shut Case
  4. More Evidence for AD Starting Without Brain Amyloid
  5. Scientists Propose a New Definition for Tau-Only Pathology
  6. Memory Concerns Presage Cognitive Decline and Aβ Pathology
  7. Controlling Blood Pressure May Lower Amyloid in ApoE4 Carriers
  8. Immunotherapy I: Baby Steps, but No Breakthroughs

Paper Citations

  1. . Cognitive trajectories associated with β-amyloid deposition in the oldest-old without dementia. Neurology. 2013 Apr 9;80(15):1378-84. PubMed.
  2. . Effect of amyloid on memory and non-memory decline from preclinical to clinical Alzheimer's disease. Brain. 2014 Jan;137(Pt 1):221-31. Epub 2013 Oct 30 PubMed.
  3. . Aβ and cognitive change: Examining the preclinical and prodromal stages of Alzheimer's disease. Alzheimers Dement. 2014 Feb 28; PubMed.
  4. . Amyloid-β related memory decline is not associated with subjective or informant rated cognitive impairment in healthy adults. J Alzheimers Dis. 2015;43(2):677-86. PubMed.

Further Reading