Tau PET Misses Early AD Pathology, Primary Age-Related Tauopathy

Tau PET has become a valuable research tool, but questions remain about how well tracer uptake reflects underlying tangles, how plaques influence this relationship, and how effectively the scans could be used for diagnosis. In the July 24 Science Translational Medicine, scientists led by Keith Josephs at the Mayo Clinic in Rochester, Minnesota, examined these questions in the largest study yet to correlate tau scans during life with postmortem autopsy data from the same participants.

  • Flortaucipir uptake is detectable only when there is a high plaque and tangle load.
  • The more plaques a person has, the more sensitive flortaucipir is to tangles.
  • Flortaucipir cannot distinguish people with PART from controls.

In a cohort of 248 people, the tau tracer flortaucipir detected tangles in Braak stage V or VI brains, but gave little signal at earlier stages. Braak staging describes the spread of tangles from the entorhinal region (I and II) to the medial temporal lobe (III and IV) and finally the cortex (V and VI).

That said, the presence of amyloid plaques boosted the tau signal. In people whose plaque burden was high, flortaucipir also lit up Braak stage III or IV. In effect, tau scans were able to identify people with overall high levels of Alzheimer’s disease neuropathologic change, but not those with low levels.

The findings have implications for the clinical use of tau PET, since a negative scan does not necessarily mean a person is free of pathology, noted first author Josephs. “We need to be cautious when using these molecular PET scans to determine eligibility for treatment, especially if treatment aims to target early stages of the disease,” Josephs wrote to Alzforum. For example, demanding that a patient is tau-PET-positive for them to receive donanemab might exclude some with early stage tauopathy.

The scans also failed to detect tangles in people with primary age-related tauopathy (PART). “The inability of flortaucipir to detect PART is a very important observation, because the field has been misinformed by several in vivo PET studies that have used inappropriate thresholds,” noted Rik Ossenkoppele at Vrije University Amsterdam. He suggested that studies test whether other tau tracers, such as MK-6240, might do a better job of this (comment below).

Tau PET Misses Early Disease. The flortaucipir signal in the entorhinal cortex (y axis) is elevated only in people with high AD neuropathologic change, not in those with low to intermediate pathology or with only tangles (PART). Pink circles indicate positive PiB scans; green negative; gray, no PiB scan. SUVR increments between 1.0 and 1.5 are represented by different shades. [Courtesy of Josephs et al., Science Translational Medicine/AAAS.]

Flortaucipir is the only tau PET tracer approved by the U.S. Food and Drug Administration. Previous autopsy studies had shown that it can identify AD patients at Braak stage V or VI, but how well it works at earlier stages was unclear (May 2020 news).

Josephs and colleagues investigated this in participants from the Mayo Clinic Study of Aging, the Mayo Clinic ADRC, and the Neurodegenerative Research Group. Almost all participants were non-Hispanic whites. They were around 75 years old at the time of their tau PET scan, and came to autopsy an average of two years later. All but nine had PiB amyloid scans as well. At autopsy, three-quarters of the cohort had AD pathology; 20 percent had PART, defined as tangle pathology of Braak stage I-IV in the absence of amyloid plaques (Aug 2023 news). Five percent had no Alzheimer’s pathology. The authors compared these neuropathologic findings with amyloid and tau PET in each person.

To the authors’ surprise, flortaucipir binding was more sensitive in people with high amyloid loads. For those with the highest plaque burden, i.e., Thal Phase 5, a flortaucipir signal could be detected in the medial temporal lobe (MTL), namely the entorhinal cortex and hippocampus, when tangles were confined to those regions, i.e., Braak stage II to III. For those at Thal Phase 3 or 4, tangles were only detectable in these MTL regions at Braak stage III or higher. When there were few plaques, i.e., Thal Phase 1 or 2, MTL tangles were only detected by PET once they had spread to Braak stage IV or higher. When plaques were entirely absent, MTL tangles were only detectable after pathology had spread to the neocortex, i.e., Braak stage V and VI. In short, advanced plaque and tangle pathology somehow combined to boost the tau PET signal in the MTL. This meant that people with PART, who have no plaques and never reach Braak stage V or VI, had no flortaucipir signal.

How might plaques influence this? Because a previous Mayo Clinic study found that flortaucipir can weakly bind sparse tau deposits in neuritic plaques themselves, the authors speculated that some of the signal might come directly from plaques (Lowe et al., 2016). 

Oskar Hansson at Lund University, Sweden, considers this unlikely. He noted that a high plaque load correlates with having denser tangles in the medial temporal lobe (Feb 2020 news). This would produce a higher flortaucipir signal. Because postmortem Braak staging only examines the location of tangles, the postmortem analysis in this study did not assess their density. “It would be great if a follow-up study could compare PET and neuropathological measures of the concentration of abnormal tau,” Hansson suggested (comment below). He predicted this would show the flortaucipir signal coming only from tangles, not plaques. The authors intend to explore this.

Finally, the authors delved deeper into whether regional tau PET signals could help discriminate between PART and healthy controls. The entorhinal cortex did this best, but managed only 80 percent sensitivity and 60 percent specificity. Likewise, EC flortaucipir uptake did not reliably distinguish PART from early AD (see image above). Adding PiB PET did not help pick out AD, as many people with Thal Phase 1 or 2 were PiB-negative, but positive for EC tau. Thus, people with very early AD may be misclassified as PART, the authors cautioned.

It remains to be seen if other tau tracers might outperform flortaucipir in identifying PART. Josephs thinks none of the current crop are up to the task. “I am not aware of any other β-amyloid or tau tracers that have been shown to be more sensitive to lower levels of AD pathology using autopsy-confirmed data. More sensitive PET tracers are clearly needed,” he told Alzforum.

Meanwhile, Peter Nelson at the University of Kentucky, Lexington, questioned what role PET scans will play in routine dementia diagnostics. “Will all patients undergo multiple batches of expensive radioactive PET scans? It seems that, instead, biomarker technologies are moving toward the analyses of blood, which may greatly 'democratize' the assessment of A/T/N biomarkers and beyond,” Nelson wrote to Alzforum.—Madolyn Bowman Rogers

Comments

  1. This is an excellent and important study reporting on the largest antemortem tau PET (combined with amyloid PET) versus autopsy dataset to date. The main goal of the study was to investigate the relationships between tau and amyloid pathology based on information derived from in vivo and postmortem data. The main strengths include i) the impressive sample size, ii) the relatively short time interval between last PET scan and death, iii) the good representation of various syndrome diagnoses, iv) the wide range of the tau/amyloid pathological burden, and v) the availability of both tau and amyloid PET in nearly all cases.

    The inability of flortaucipir to detect primary age-related tauopathy (PART) is a very important observation, because the field has been misinformed by several in vivo PET studies that have used inappropriate thresholds (i.e., too low for tau PET and too high for amyloid PET) to suggest that flortaucipir can be used to detect PART in vivo. Future antemortem versus postmortem studies will be key to test whether other tau PET tracers (especially MK6240) are suitable for the detection of PART in vivo. This is plausible given the higher affinity of MK6240 as well as the high(er) proportion of A-T2+ cognitively unimpaired individuals in independent MK6240 versus flortaucipir PET studies.

    View all comments by Rik Ossenkoppele

  2. This is a very comprehensive study, which was done in a careful manner by experts in the field. I have only one major concern, and that is how A plaques and tau tangle pathology are compared to flortaucipir retention. The flortaucipir PET signal is quantified in nine different regions of interest in a semiquantitative manner, as we normally do, and these SUVR-based measures should reflect the density of tau aggregates in the same nine regions. However, the neuropathological assessments are done using the standard neuropathological staging systems, Braak and Thal, which only tell us where in the brain tau aggregates and A plaques, respectively, can be found.

    Of course, there is an association between Braak stage and the regional tau tangle density in different brain regions, but it is naturally far from perfect. For example, people at Braak stage III-IV, without amyloid, have a lower density of tau aggregates in the medial temporal lobe than do those with a high Ab plaque load. Therefore, comparing flortaucipir retention in different brain regions with Braak (or Thal) staging is a bit complex for me to interpret.

    It would be great if a follow-up study could compare PET and neuropathological measures of the density (or concentration) of abnormal tau in the same regions. This could be done by comparing flortaucipir retention with the density of abnormal tau in the same regions as detected by immunohistochemistry, e.g., with AT8 antibody, or by quantifying the levels of abnormal tau species in fresh frozen tissue. I believe that when using both a measure of the regional density of plaques and a measure of the regional density of tau tangles and neuropil threads, only the latter will predict the regional flortaucipir retention. I do not believe that flortaucipir can detect the rather low level of tau aggregates found in Ab plaques.

    View all comments by Oskar Hansson

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Flortaucipir Autopsy Study Published
  2. Alzheimer’s, Primary Age-Related Tauopathy Take Separate Paths
  3. How Much Amyloid Will Kick Off Tangles, and Decline?

Paper Citations

  1. Lowe VJ, Curran G, Fang P, Liesinger AM, Josephs KA, Parisi JE, Kantarci K, Boeve BF, Pandey MK, Bruinsma T, Knopman DS, Jones DT, Petrucelli L, Cook CN, Graff-Radford NR, Dickson DW, Petersen RC, Jack CR Jr, Murray ME. An autoradiographic evaluation of AV-1451 Tau PET in dementia. Acta Neuropathol Commun. 2016 Jun 13;4(1):58. PubMed.

Further Reading

Primary Papers

  1. Josephs KA, Tosakulwong N, Weigand SD, Graff-Radford J, Schwarz CG, Senjem ML, Machulda MM, Kantarci K, Knopman DS, Nguyen A, Reichard RR, Dickson DW, Petersen RC, Lowe VJ, Jack CR Jr, Whitwell JL. Flortaucipir PET uncovers relationships between tau and amyloid-β in primary age-related tauopathy and Alzheimer's disease. Sci Transl Med. 2024 Jul 24;16(757):eado8076. PubMed.

Annotate

To make an annotation you must Login or Register.