In one of the first multimodal imaging studies of its kind, scientists have confirmed in vivo what they could once only determine at autopsy—a tight correlation between tau neurofibrillary tangles and neurodegeneration in individual patients in early clinical stages of various forms of Alzheimer’s disease. The work, from Brad Dickerson and colleagues at Massachusetts General Hospital in Boston, was published online February 20 in JAMA Neurology.

As expected, the regional distribution of tau tangles differed among patients with typical and atypical forms of AD, but in every case, atrophy co-localized with tau, and did so in brain regions linked to specific clinical symptoms, such as memory impairment, difficulties with language, or visual or motor problems. In contrast, no significant correlation emerged between amyloid PET signals and either atrophy or symptoms. 

Tau Predicts Atrophy Predicts Symptoms. In a patient with corticobasal syndrome, tau PET tracer uptake (left panel) coincides with cortical atrophy as per quantitative MRI (center panel), but not with amyloid tracer uptake (right). [© 2017 American Medical Association. All rights reserved.]

The results provide further in vivo evidence for what decades of autopsy data have indicated—that the location of tau in the brain, rather than amyloid, dictates symptoms. “Pathologists have been telling us this for years,” said Dickerson. “But they were looking at postmortem brain tissue after a person had died at an advanced stage, so most of us were still wondering what we would see in a person at a relatively mild stage of illness.” 

Dickerson pointed out that researchers led by Rik Ossenkoppele, then working at Gil Rabinovici’s lab at the University of California, San Francisco, were the first to link tau to reduced brain function in a living person, who had posterior cortical atrophy (Ossenkoppele et al., 2015Feb 2015 conference news). “That was a real milestone for the field,” he said. “I’m happy we can confirm and extend those findings.”

“It's a relatively small number of subjects, but the relationships between tau and atrophy are striking,” said Rabinovici, who was not involved in the new study. “Using multiple imaging technologies in individual patients, and seeing a common theme of that tight link between tau, structural MRI, and clinical presentation, along with the dissociation from amyloid deposition, at least in this stage, is really informative.”  

Postmortem studies of AD brain tissue proved decades ago that clinical state and neuronal loss track with the presence of neurofibrillary tangles and not Aβ plaques. The advent of amyloid and tau PET ligands allowed researchers to obtain a much broader look at both pathologies in living subjects over the course of disease. Key studies in the past year have shown that tau PET can identify disease progression and pathological staging similar to traditional Braak criteria (see Mar 2016 newsMay 2016 news; Johnson et al., 2016). In addition, combined imaging studies have shown that, at least on the group level, regional distribution of tau, and not amyloid, tracks best with neurodegeneration measured by either brain volume or neuronal function (see Aug 2016 news; Ossenkoppele et al., 2016). 

In the new study, first author Chenjie Xia and colleagues imaged six dementia patients. Three had typical amnestic AD, and three had an atypical variant—either posterior cortical atrophy (PCA), logopenic variant primary progressive aphasia (lvPPA), or corticobasal syndrome (CBS). They underwent PET imaging with [18F]AV-1451 for tau and [11C]PiB for amyloid, and structural MRI to measure cortical thickness.

In the patients with typical AD, tau PET revealed bilateral uptake of the tracer in brain regions classically associated with memory impairment in AD, including the posterior cingulate, precuneous, lateral temporoparietal, and occipital cortices. Whole brain mapping of cortical thinning, a correlate of atrophy, closely paralleled tau tracer uptake in all three patients.

In the patients with atypical AD, the tau picture was different, and unique for each clinical syndrome. In PCA and lvPPA, for example, tau accumulated in visual and language association centers, respectively. In corticobasal syndrome, tau appeared in the sensorimotor cortex, consistent with a clinical phenotype of motor impairment, and mostly on the right side of the brain, reflecting the patient’s predominantly left-sided symptoms (see image above). Just as with typical AD cases, the regions of cortical thinning coincided nearly completely with areas of tau: Most cortical areas with definite atrophy had a substantial tau burden, while very few regions with low tau showed signs of atrophy.

In contrast, amyloid distribution in all six cases was fairly similar and correlated poorly with areas of neurodegeneration. A regression analysis comparing the magnitude of tau or amyloid signals with degree of atrophy in 68 regions of interest across the brain further revealed that the tau signal correlated with the extent of atrophy, while amyloid load did not.

In light of the tight link between tau and neurodegeneration, could tau PET function as a biomarker for both? Perhaps, said Ossenkoppele, now at VU University Medical Center, Amsterdam, the Netherlands. “The tau PET pattern reflects binding to tau pathology, but there is a huge spatial overlap with neurodegeneration—nearly all of the regions with high tau also show reduced cortical thickness. I think it could be a really efficient tool to get an impression of both the pathology and downstream neurodegeneration.”

However, poor specificity for tangles remains a major cautionary point about AV-1451 (see Feb 2016 conference news). “We’re still not entirely certain we are only visualizing paired helical filaments. There may be other non-specific binding or even age-related, non-AD tau pathology that is contributing to the tau PET signal,” said Val Lowe, Mayo Clinic, Rochester, Minnesota. The pathology will need to be confirmed by autopsy as time goes on, he said.

The study suggests a possible means to more definitively diagnose atypical types of AD, although many more patients need to be studied before that can occur, said Dickerson. In addition, he is looking to do longitudinal studies to understand what happens to tau and atrophy over time as disease progresses.

Variant AD will be an important challenge for the field to tackle going forward, said Rabinovici. “These presentations aren’t that rare. They account for around 15-20 percent of AD patients who are seen in specialized dementia clinics,” he said. “What we’re calling atypical AD may not be that atypical after all.”

Patients with non-amnestic dementia tend to be younger, do not have an ApoE4 allele, and may not be immediately recognized as having AD. For these reasons, they may not have access to clinical trials. “The current diagnostic criteria recognize these variants as different types of AD, and with biomarker evidence these people could be more easily diagnosed,” said Rabinovici. “They should have an opportunity to participate and have access to therapeutic drugs like older, more amnestic patients.” He told Alzforum that he, Dickerson, and Liana Apostolova of Indiana University in Indianapolis are collaborating to try to establish a multisite U.S. network to register atypical AD patients for further studies, which they hope will facilitate trials targeting these patients, or at least including them.

The results also offer more support for the idea that tau could be a key target for preventing dementia or modifying disease course. As the authors write, “This and other tau ligands of PET imaging will likely add critical momentum to the design of next-generation clinical trials targeting tau pathologic findings.”

The study has important implications for treatment, Rabinovici added. “Preventing the spread of tau is going to be critical to modify the course of the disease and help prevent clinical decline. Whether amyloid-lowering drugs will be sufficient to do this or whether we need to hit tau or the links between Aβ and tau remains an open question, but tau is the smoking gun here.”—Pat McCaffrey

Pat McCaffrey is a writer based in Newton, Massachusetts.


  1. This is the first study to report the relationship of cortical thinning assessed with structural MRI and hyperphosphorylated tau pathology measured with [18F]AV-1451. The main findings, namely the differential patterns of AV-1451 uptake in AD variants and the strong spatial overlap of cortical atrophy and brain regions affected by tau pathology, are confirming previous results (e.g. Ossenkoppele et al., 2016) where neuronal dysfunction/neurodegeneration was assessed using FDG PET rather than MRI.

    The study generally strongly confirms previous findings from both neuropathology and neuroimaging studies, strengthening the knowledge that tau pathology is closely related to measures of neurodegeneration—more closely than is Aβ pathology.

    Despite the small sample sizes in most published studies, the reproducibility of findings using tau PET is striking and sends a very encouraging signal to the biomarker community. It is noteworthy, however, that all patients in the current study, including the amnesic AD patients, are rather young, representing early onset AD, and we have found that these  differ from late-onset AD both in AV1451 uptake and cortical atrophy pattern.

    One finding that is not commented on in detail is that regions affected by tau pathology generally seem larger than atrophic regions, raising the question whether tau pathology occurs before neurons become dysfunctional or perish.

     It is obviously difficult to draw strong conclusions from such a small sample but the present findings, combined with the knowledge gleaned especially during the past three years, support the idea that in vivo tau burden, as seen in these scans, is a valid marker for clinical symptoms and neurodegeneration in AD.

    With obvious interplay between amyloid and tau pathology, the latter seems to be driven by the former, so it is not one or the other that is most important, but rather knowing when and where each pathology occurs. In this regard, PET scans are, and will continue to be, extremely valuable tools for both identifying suitable candidates for treatment trials and as outcome measures. 


    . Tau PET patterns mirror clinical and neuroanatomical variability in Alzheimer's disease. Brain. 2016 May;139(Pt 5):1551-67. Epub 2016 Mar 8 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. What If It’s Not Garden-Variety AD? Telling Variants Apart by Where Tau Is
  2. Tau PET Aligns Spread of Pathology with Alzheimer’s Staging
  3. On Multiple Marker Analysis, Tangles Track Best With Functional Decline
  4. Brain Imaging Suggests Aβ Unleashes the Deadly Side of Tau
  5. Shaky Specificity of Tau PET Ligands Stokes Debate at HAI

Paper Citations

  1. . Tau, amyloid, and hypometabolism in a patient with posterior cortical atrophy. Ann Neurol. 2015 Feb;77(2):338-42. Epub 2014 Dec 17 PubMed.
  2. . Tau positron emission tomographic imaging in aging and early Alzheimer disease. Ann Neurol. 2016 Jan;79(1):110-9. Epub 2015 Dec 15 PubMed.
  3. . Tau PET patterns mirror clinical and neuroanatomical variability in Alzheimer's disease. Brain. 2016 May;139(Pt 5):1551-67. Epub 2016 Mar 8 PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Association of In Vivo [18F]AV-1451 Tau PET Imaging Results With Cortical Atrophy and Symptoms in Typical and Atypical Alzheimer Disease. JAMA Neurol. 2017 Apr 1;74(4):427-436. PubMed.
  2. . Tau and Cortical Thickness in Alzheimer Disease. JAMA Neurol. 2017 Apr 1;74(4):390-392. PubMed.