By the time cognitive symptoms of Alzheimer’s disease manifest, PET scans reveal rampant amyloid throughout the brain. A new study tracks the regional progression of this hallmark pathology from beginning to end. In the October 18 Neurology, researchers led by Michel Grothe of the German Center for Neurodegenerative Diseases in Rostock report that they have developed a four-stage model of regional Aβ deposition. Drawing on imaging data from nearly 700 volunteers in the Alzheimer’s Disease Neuroimaging Initiative (ADNI), they determined that the spatial distribution of Aβ plaques changed in a consistent manner as disease progressed, first appearing in the basal part of the temporal lobe, the anterior cingulate gyrus, and the bottom of the parietal lobe, then sweeping throughout the neocortex and eventually inward, overtaking subcortical regions such as the striatum.

  • Florbetapir PET reveals a four-stage model of regional amyloid deposition.
  • Scans from 98 percent of people along the AD spectrum fit the scheme.
  • Stages correlated with CSF Aβ42, clinical disease stage, and memory scores.

The pattern aligns remarkably well with findings from postmortem neuropathological studies. PET scans from 98 percent of volunteers with detectable Aβ fit along the staging spectrum. Notably, regional staging detected Aβ accumulation in many cognitively normal people who would have been considered amyloid-negative using commonly applied global PET tracer thresholds.

“This study opens exciting new avenues for future research, and further points to the pressing question of what it means to have amyloid in the brain, especially in the earliest stages of Aβ accumulation,” wrote Gaël Chételat of the University of Caen-Normandie in France and Melissa Murray of the Mayo Clinic in Jacksonville, Florida, in an accompanying editorial. They added that longitudinal data will be crucial to validate the staging model.

Using postmortem neuropathology, researchers previously established that Aβ pathology spreads throughout the brain in a characteristic pattern. Fifteen years ago, Dietmar Thal, then at Goethe University in Frankfurt, Germany, examined nearly 50 brains to derive five stages of Aβ deposition—moving from the neocortex to allocortex, and eventually inward to the brainstem and cerebellum (Thal et al., 2002). People in the earlier stages tended to be cognitively normal when they died, while those in later stages were likely to have been clinically diagnosed with AD.

Regional Frequencies.

Among cognitively normal people, the frequencies of Aβ positivity, ranging from zero (blue/black) to 55 percent (red), were calculated for each of 52 brain regions (left, shown in three cross-sectional views). This continuum of regional frequencies was then equally partitioned into four stages (right). [Courtesy of Grothe et al., Neurology, 2017.]

Such a regional staging scheme has not yet been recapitulated using live imaging, however. For practical purposes, an amyloid-PET scan is considered positive if the global cortical standard uptake value ratio (SUVR), typically in comparison to uptake in the cerebellum, breaches a predefined threshold—usually 1.17. By the time cognitive symptoms manifest, most people have already been “Aβ positive” by this global measure for some time; on the flip side, cognitively normal people with positive scans often have inklings of memory problems (Jack et al., 2010; Bennet et al., 2012).

Grothe and colleagues set out to develop a regional Aβ staging scheme based on florbetapir-PET scans. They hypothesized that distinct changes in the distribution of Aβ would correlate with disease progression. Because most Aβ deposits before cognitive symptoms emerge, the researchers used PET scans from 179 cognitively normal people in the ADNI cohort. For each of 52 brain regions, the researchers calculated the percentage of cognitively normal participants who had an SUVR greater than 1.17. Next they used the frequency of Aβ positivity as an indicator of regional progression, with the most frequently affected areas being indicative of earliest stage and least frequently positive areas being stage IV. This staging method, commonly used in neuropathology, assumes that as a pathological burden increases, the number of people surviving with that burden or, in this case, having no cognitive symptoms at the time of their scan, decreases.

They found that roughly half of the cognitively normal people had Aβ deposition in various regions of the neocortex, including the basal part of the temporal lobe, the anterior cingulate, and the parietal operculum—an inner region of the parietal cortex that covers the insular cortex. Broader regions of the temporal, frontal, and parietal lobes harbored Aβ plaques in 30–40 percent of volunteers, while only about 15–25 percent had deposits in the sensory motor cortices and anterior lobe structures. Finally, only 5–10 percent accumulated Aβ in the striatum. 

The researchers then grouped brain regions by highest to lowest frequency of Aβ positivity into four anatomic divisions, which they proposed would correlate with stages of progression of Aβ pathology. These stages coincided with those developed independently from Thal’s and other neuropathology studies.

Amyloid Stages. Regional progression of Aβ deposition, as proposed by the four-stage model. Each progressive stage includes new affected regions (red) as well as affected regions from the previous stage (blue). [Courtesy of Grothe et al., Neurology, 2017.]

To validate their model, the researchers attempted to fit a broader set of scans into their staging scheme. From 667 ADNI volunteers who had diagnoses ranging from normal to AD, 418 had evidence of Aβ accumulation, and 410 of those (98 percent) had regional deposition patterns that fit squarely into one of the four stages. The striking inclusivity of the model surprised Grothe, especially given that it was generated without making any a priori assumptions of regional progression patterns. Michael Donohue of the University of Southern California in San Diego told Alzforum that he was also surprised by how few participants failed to conform to the staging model. 

How would this four-stage model compare to the dichotomous global SUVR cutoff tests for amyloid? The researchers found that all stage IV and most stage III participants tested positive by global PET. However, about half of stage II and the vast majority of stage I tested negative. Even after reducing the global threshold from 1.17 to 1.10, 25 percent of stage II and 81 percent of stage I volunteers tested negative. The findings suggest that regional staging catches people in earlier stages of amyloidosis than the standard global test does. However, Grothe pointed out that the clinical significance of these early stages is still unclear.

The researchers next compared the concentration of cerebrospinal fluid Aβ42 with the regional Aβ stages, and found that CSF Aβ42 dropped continuously across progressive amyloid stages, with even stage I having lower concentrations than people without evidence of Aβ deposition, who were considered “stage zero.” This suggests that early regional deposition, which would have been missed via global measures, correlates with other biomarkers and has physiological meaning, Grothe said.

Henrik Zetterberg of the University of Gothenburg in Sweden was fascinated by the finding that Aβ deposition in quite restricted brain regions was sufficient to have an impact on lumbar CSF Aβ concentrations. He added that this could explain why among cognitively normal people and some with MCI, global amyloid PET scans come up negative while CSF Aβ tests are positive.

The regional Aβ stages also correlated with clinical diagnosis. Ninety-six percent of patients with Alzheimer’s dementia fell into stage III or IV, compared with 66 percent of people with MCI and 42 percent of cognitively normal volunteers. Higher amyloid stages correlated with poorer performance on memory tests, but only for cognitively normal people or those with MCI, not AD, who scored lowest in these tests. In contrast, testing positive for global Aβ correlated with worse memory scores in the MCI and AD groups, but not in cognitively normal people. Together, these findings suggest that the regional Aβ staging scheme correlates with the severity of early cognitive symptoms, while the global measures correlate only with more severe cognitive decline in later clinical stages of disease.

“The staging suggests the possibility of a more inclusive and/or sensitive approach to screening for clinical trials in preclinical or prodromal populations,” Donohue wrote. “As part of the validation, it will be interesting to examine disease progression for individuals identified as stage I or II but not reaching conventional global SUVR thresholds.” Clinical trials that use global amyloid positivity by PET as an inclusion criterion could be excluding people with sub-threshold scans even though they are in fact depositing considerable brain amyloid and are on the path to AD.

Grothe emphasized that the clinical significance of early stage Aβ deposition remains unclear. The current definition of preclinical AD—a positive global Aβ scan in the absence of cognitive impairment—already includes about 20 percent of people over 70, Grothe said. Including people who fall into amyloid stage I or II of the regional scheme could bump that number up to 40–50 percent. Interestingly, while high, this number aligns more closely with postmortem neuropathological findings, which have found around half of cognitively normal elderly adults harbor Aβ deposits.

Grothe acknowledged the possibility that this early regional deposition could, in theory, reflect a benign aging process. Longitudinal data will show if or how fast people in early stages progress to later stages and clinical symptoms, he said. This makes using regional measures for clinical trial selection a dicey undertaking for now. The researchers plan to incorporate longitudinal data into the scheme. Recent longitudinal data from ADNI and the Australian Imaging, Biomarkers, and Lifestyle study have reported that Aβ positivity predicted eventual cognitive decline, at least when global Aβ PET or CSF Aβ42 were used (Jun 2017 newsAyton et al., 2017Harrington et al., 2017). Grothe will also assess how various risk factors, including cardiovascular health, diet, and exercise, influence the rate of progression.

Grothe added that if validated, regional amyloid staging could be used in combination with tau PET scans, which have been developed with regional staging in mind from the get-go. He wondered whether stage I/II of Aβ deposition would be sufficient to drive tau out of the temporal lobe, as has been demonstrated for global Aβ measures.—Jessica Shugart


  1. This is an interesting study by Grothe and colleagues that suggests a typical pattern of regional deposition for neocortical Aβ in AD. Interestingly, the staging criteria they apply to the regional deposition allows for more than 45 percent of older clinically normal people to be identified as having unusual amyloid deposition, opposed to the ~30 percent identified using traditional dichotomous categorizations. This could have implications for clinical trials, allowing individuals to be identified at even earlier stages of the disease pathogenesis. However, caution must be taken as these findings are based on cross-sectional data. It will be important to validate these findings, particularly by understanding the rates of progression through the stages identified as well as rates of progression towards disease.

  2. In this staging scheme proposed by the authors, the consistency of the pattern is certainly striking and will be very useful in terms of recruitment into clinical trials and diagnostic prognosis. What strikes me is the discordance between very early pre-symptomatic appearance of striatal Aβ in autosomal-dominant AD (ADAD), as shown by myself, Bill Klunk, and Victor Villemagne years ago, and the very late appearance in sporadic disease, something Grothe and colleagues also point out in their discussion. This is certainly something to think about when using ADAD as a model for Alzheimer’s disease in general.

  3. In this experiment, Grothe and colleagues find evidence for in vivo staging of amyloid based on regional uptake values extracted from florbetapir PET data from the ADNI. To do this, the authors examine regional positivity within the clinically normal sample, and determine the frequency by which this positivity occurs. For instance, the anterior cingulate and fusiform gyrus are in the earliest stage since they are positive in the most individuals; the striatum is in the final stage since the prevalence of positivity is lowest in that region. The overall result is a four-stage model of amyloid accumulation that for the most part recapitulates postmortem Thal staging. However, the results suggest finer resolution than Thal Stage 1 by implying specific regions within the neocortex (inferior/basal temporal and anterior cingulate). In my view, the “Grothe Stage 1” result is the most important finding in this paper. It has long been observed that amyloid deposition observed with PET imaging, even among normals, shows a widely distributed regional pattern that impacts multiple cortical regions. This work contradicts that dogma by suggesting a subset of regions show early amyloid accumulation. Importantly, Grothe Stage 1 already shows reduced CSF amyloid, suggesting that this staging procedure may be able to capture very early amyloid pathology that is regionally restricted. The consequences of early deposition in the inferior/basal temporal lobe and anterior cingulate among clinically normal individuals remains unknown, and may provide important insights for the early consequences on cognition and neuronal integrity in older normal individuals. Interestingly, an initial site of tau deposition within amyloid-positive normals seems to be in the inferior temporal cortex (Schöll et al., 2016), suggesting there may be some unique vulnerability in this region. Furthermore, over the past few years we have witnessed a shift in clinical trial design, such that many strategies are currently underway to target asymptomatic individuals with evidence of abnormal amyloid as early as possible. As shown in this article, traditional global cut offs are missing 81 percent of Grothe Stage 1. Thus, implementation of a more regionally specific mask may enable the detection of individuals who are at the earliest stages of amyloid accumulation for enrollment into prevention trials.

    A few new directions can be suggested based on the results of this work. As mentioned by the authors, it may be important to consider region-specific cut-offs, given that differences in gray-matter density and surface area may influence the PET signal extracted across regions; thus a general cut-off may not be appropriate. The fact that Grothe Stage 1 already showed reduced CSF amyloid suggests that the involvement of these regions is not purely driven by cut-off selection. It would be interesting to apply this approach to young APOE4-positive subjects as a “control” to ensure that these gold-standard amyloid cases would fall into Grothe Stage 0. Finally, it would be informative to see these patterns longitudinally, given that the thrust of this work is to provide a sequence of involvement across amyloid stages. If Stage 1 becomes Stage 2 over time, then that would provide convincing evidence for Grothe Staging.


    . PET Imaging of Tau Deposition in the Aging Human Brain. Neuron. 2016 Mar 2;89(5):971-82. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. At Risk, or Already Alzheimer’s? Elevated Aβ Predicts Cognitive Decline

Paper Citations

  1. . Phases of A beta-deposition in the human brain and its relevance for the development of AD. Neurology. 2002 Jun 25;58(12):1791-800. PubMed.
  2. . Hypothetical model of dynamic biomarkers of the Alzheimer's pathological cascade. Lancet Neurol. 2010 Jan;9(1):119-28. PubMed.
  3. . Relation of neuropathology to cognition in persons without cognitive impairment. Ann Neurol. 2012 Oct;72(4):599-609. PubMed.
  4. . Cerebral quantitative susceptibility mapping predicts amyloid-β-related cognitive decline. Brain. 2017 Aug 1;140(8):2112-2119. PubMed.
  5. . Amyloid β-associated cognitive decline in the absence of clinical disease progression and systemic illness. Alzheimers Dement (Amst). 2017;8:156-164. Epub 2017 Jun 9 PubMed.

Further Reading

Primary Papers

  1. . In vivo staging of regional amyloid deposition. Neurology. 2017 Nov 14;89(20):2031-2038. Epub 2017 Oct 18 PubMed.