Tau PET Best Predicts Short-Term Decline in Early Alzheimer’s

People with Alzheimer's need a prognosis. When, and how fast, will they get worse? To help clinicians decide which biomarker most accurately anticipates waning cognition in early Alzheimer’s, researchers led Oskar Hansson at Sweden's Lund University compared amyloid PET, tau PET, and structural MRI side by side. In the June 28 JAMA Neurology, they declared tau PET the winner, at least when cognition was approximated with a simple screen over a two-year time span. In 1,431 older adults spanning from cognitively normal to AD dementia, tangle load predicted falling mini-mental state exam scores better than did plaque load or cortical thickness.

  • Tau PET outperformed amyloid PET, MRI to predict change on MMSE.
  • At a given tau load, older people declined faster than younger ones.
  • In contrast, sex and APOE genotype made no difference in this study.

Among the handful of most-studied AD biomarkers, tangles track most closely with cognitive decline (Jan 2020 conference news). In recent work, tau PET stood out as the most accurate imaging technique for diagnosing AD, distinguishing it from other dementias with greater than 90 percent accuracy (May 2020 news; Sep 2018 news). But how good is tau PET at predicting AD symptoms?

To find out, first author Rik Ossenkoppele and colleagues gathered flortaucipir tau and amyloid PET scans, MRI scans, and MMSE scores from 1,135 people from across seven cohorts in the U.S., Sweden, and South Korea. The cohorts are the Alzheimer's Disease Neuroimaging Initiative, the Berkeley Aging Cohort Study, one by Avid Radiopharmaceuticals in Philadelphia, Lilly’s Expedition-3 solanezumab study placebo arm, Swedish BioFINDER-1, and cohorts at the University of California in San Francisco and at the Korean Memory Disorder Clinic of Gangnam Severance Hospital. For a replication sample, the researchers collected the same imaging and cognition data from 296 participants in the Swedish BioFINDER-2 cohort, which uses the RO948 tracer for tau PET scans. Because each of these cohorts was designed individually, and not with data pooling in mind, the crude but ubiquitous MMSE screen was the only cognitive measure common to all, i.e., available for this analysis, the authors acknowledge.

Of all participants, 315 had AD dementia, 271 had Aβ-positive MCI, 172 had Aβ-negative MCI, 253 were Aβ-positive but cognitively normal, and 420 were Aβ-negative and cognitively normal. Fifty-two percent were men; the average age was 71. Amyloid scans were done within six months of tau scans, and at least two MMSE scores within an average of two years were recorded for each participant.

Imaging Versus Cognition. Tau tangles (left) correlated more tightly with cognitive decline than did amyloid plaques (middle) or thinning in AD-specific cortical regions (right). [Courtesy of Ossenkoppele et al., JAMA Neurology, 2021. © 2021 American Medical Association. All rights reserved.]

How did the techniques compare? For tau versus amyloid PET, both flortaucipir and RO948 uptake into the entorhinal cortex tracked more closely with annual cognitive decline than did plaque load. This was the case in all participants, except for people who had Aβ-negative MCI. For tau PET versus MRI, in the discovery cohort, greater flortaucipir uptake tracked more tightly with MMSE worsening than did cortical thickness in the AD-signature region—the bilateral entorhinal, inferior, middle temporal, and fusiform cortices. This was true in Aβ-positive MCI and in all cognitively normal participants. The reason could be that person-to-person variation in cortical thickness obscures AD-related shrinkage, the authors suggest. In contrast, MRI outperformed tau PET in the AD dementia and Aβ-negative MCI groups. In the replication cohort, RO948 uptake prevailed in all the same groups as flortaucipir and also in people with AD dementia, possibly reflecting that tracer's wider dynamic range, the authors write. MRI bested tau PET in people with Aβ-negative MCI.

Chris Rowe, University of Melbourne, Australia, cautioned that most cognitively unimpaired participants had negative tau PET scans. This fits an emerging consensus that tau PET becomes positive brain-wide only once a person enters the symptomatic stage of AD—and precludes much predictive power for tau PET at the presymptomatic stage of AD (full comment below).

Combining techniques improved the prediction, but not equally so. When tau PET was added to an amyloid PET or MRI scan, the tau scan offered more information than either scan provided by itself. However, amyloid PET or MRI scans added little information to what a tau PET scan alone already provided. This can inform ordering decisions in the clinic. “If a clinician only has an MRI, ordering a tau PET scan would add a lot more information than the other way around,” Ossenkoppele said.

Did demographics affect decline? Comparing tau PET scans to MMSE scores in people grouped by age, sex, or APOE, the researchers noticed that cognition slid faster in people over 70 than under 70 who had similar tau loads (see image below). This might be because with advancing age, vascular changes and other proteinopathies such as TDP-43 exacerbate neurodegeneration, the researchers write. Neither sex nor APOE genotype changed the rate of cognitive decline at equal tau loads. None of the three factors led to a discernable difference in the smaller replication cohort. “I feel like if we had had a bigger sample size, we would have found a significant age effect,” Ossenkoppele said.

Worse With Age. In people with comparable tangle loads (x axis), older age (left) hastened cognitive decline (y axis), whereas sex (middle) and APOE status (right) had no effect. [Courtesy of Ossenkoppele et al., JAMA Neurology, 2021. © 2021 American Medical Association. All rights reserved.]

Next up, the scientists plan to compare imaging to cognitive change over time at an individual rather than group level to see, once again, which marker best supports a prognosis. Bernard Hanseeuw, Massachusetts General Hospital, wondered if longitudinal changes in a person’s tangle accumulation pattern could predict decline in specific cognitive domains. His thought is inspired by Hansson and colleagues’ recent cross-sectional study defining four subtypes of AD by linking regional tangle deposition to specific cognitive deficits (Apr 2021 news). “It would be helpful to tell patients they are at risk of a particular cognitive ability declining so they could actively work on that area,” Hanseeuw told Alzforum.

“Although tau PET is already useful for clinical diagnosis, it needs to be further established as prognostic before it can be used as such in the clinic,” Ossenkoppele said. Hansson added, “It will be more useful when there are drugs on the market to help decide who should get what treatment—right now the cost/benefit ratio is too low.” Tobey Betthauser, University of Wisconsin-Madison, takes this view, as well. “As these results become available, it will be important to consider the trade-offs between affordability, access, and prediction accuracy for various biomarker modalities,” he wrote (full comment below).

Regarding affordability and access, are fluid tau markers as prognostic as PET? Several blood p-tau species recently have been found to pick up pre-symptomatic AD and track disease progression (Apr 2021 conference news; Feb 2021 news; Jan 2021 news; Mar 2020 news). The Lund group, and others elsewhere, are now directly comparing tau PET to CSF and plasma p-tau217, and also p-tau181. Blood tests will be needed for rolling out tau-based prognostic testing widely and affordably, but Hanseeuw suspects tau PET will remain superior in some ways. “PET shows the cumulative tau burden and regional load, while fluid biomarkers do not,” he said.—Chelsea Weidman Burke

Comments

  1. This is an interesting paper with a large cohort size and two-year cognitive follow-up. The paper provides further evidence that the drivers of cognitive decline change over the course of AD, with tau burden and atrophy strongly associated with the rate of decline in the MCI and dementia phases of AD while the impact of amyloid burden declines when persons become cognitively impaired.

    Clearly a positive tau PET is a good predictor of near-term clinical progression measured by decline in MMSE score. However, the value of tau PET versus other modalities for predicting decline in cognitively unimpaired (CU) persons is debatable, especially when looking longer term. Few CUs have a positive tau PET scan for binding beyond the entorhinal region, and a positive scan is rarely seen before a substantial rise in amyloid to at least 40-50 centiloids (Doré et al., 2021). 

    The current manuscript Figure 1, line 1, Parts E and F, shows that only a handful of the 667 CUs have an increase in tau tracer binding. Tau accumulation is a late event in preclinical AD. Consequently, recruitment for preclinical AD trials would be very difficult if a positive tau PET was required.

    The Australian Imaging, Biomarkers and Lifestyle study of aging (AIBL), and many other studies using composites of cognitive tests that are sensitive for detecting subtle decline in the earliest stages of AD, have shown that amyloid PET is a very good predictor of decline in the cognitively unimpaired, and the risk is related to the amount of amyloid present (van der Kall et al., 2021). For trials that aim to recruit the earliest phases of preclinical AD, amyloid measures remain the preferred method of participant selection. 

    References:

    Doré V, Krishnadas N, Bourgeat P, Huang K, Li S, Burnham S, Masters CL, Fripp J, Villemagne VL, Rowe CC. Relationship between amyloid and tau levels and its impact on tau spreading. Eur J Nucl Med Mol Imaging. 2021 Jul;48(7):2225-2232. Epub 2021 Jan 26 PubMed.

    van der Kall LM, Truong T, Burnham SC, Doré V, Mulligan RS, Bozinovski S, Lamb F, Bourgeat P, Fripp J, Schultz S, Lim YY, Laws SM, Ames D, Fowler C, Rainey-Smith SR, Martins RN, Salvado O, Robertson J, Maruff P, Masters CL, Villemagne VL, Rowe CC. Association of β-Amyloid Level, Clinical Progression, and Longitudinal Cognitive Change in Normal Older Individuals. Neurology. 2021 Feb 2;96(5):e662-e670. Epub 2020 Nov 12 PubMed.

    View all comments by Christopher Rowe
    • User Profile Image Tobey Betthauser
      University of Wisconsin-Madison School of Medicine and Public Health
    • Posted:

    Identifying factors that predict cognitive decline and clinical progression and characterizing the rate of expected cognitive decline for combinations of these factors remain top priorities for aging and dementia research, clinical practice, and clinical trials. In this paper, Ossenkopple and colleagues investigated the utility of tau PET imaging as a prognostic marker of MMSE change across the AD continuum (A+ unimpaired, MCI, dementia) and in subgroups of the AD continuum and A- control and A- MCI groups. The study used change in MMSE score over two years following a baseline visit that included structural MRI, an Aβ biomarker (predominantly amyloid PET), and tau PET imaging in a large, multi-cohort design including over 1,400 participants, making it one of the largest studies to investigate these associations to date. Key findings were that tau PET imaging explained more variance in MMSE change across the AD continuum and within subgroups of A+ MCI and A+ unimpaired compared to amyloid PET or structural MRI.

    This addition to the literature is particularly timely considering the recent FDA approval of aducanumab (Aduhelm) in the United States, as decisions are made regarding how to identify patients who may benefit from treatment, including which diagnostic procedures will be appropriate for different patient groups.

    The findings within the AD continuum largely agree with several individual cohort studies. They also agree with cross-sectional correlates between end-of-life MMSE and neuropathological staging of neurofibrillary tangles and Aβ plaques (for review, see Nelson et al., 2012). The subgroup analyses provide additional needed information about the rate of MMSE decline in various stages of AD and, importantly, how different levels of tau pathology and structural brain changes affect the rate of MMSE decline within these stages.

    These results highlight how the magnitude of amyloid burden alone is a less-potent predictor of future MMSE decline in various stages of AD than are measures of tau and neurodegeneration. This type of characterization could be highly useful for clinical trial planning to better understand the expected cognitive course for individuals based on their baseline biomarker, demographic, and clinical profiles.

    As the authors acknowledge, even when pooling biomarker data across multiple cohorts from around the world, there is still a lack of diversity in these samples. Diversity is greatly needed to understand to what extent these findings are representative. To this point, the paper mentions this was a mostly white, non-Hispanic sample.

    Also of note, the proportion of change in MMSE variance related to tau PET and MRI differed between the main study sample and the Biofinder cohort in the A+ dementia group such that MRI explained more variance in the main study sample and tau PET explained more variance in the Biofinder sample. While there are several reasons that might explain this difference in outcomes between the two samples, these were not investigated in the current work. It will be interesting to see in future work whether these differences were driven by technical reasons (e.g., tau/amyloid tracer selection, differences in structural imaging protocols, etc.) and/or heterogeneity within the A+ dementia group between samples/studies.

    Perhaps a more subtle finding was how the association between MMSE change and tau PET differed slightly between brain regions across the different disease stages. This provides some additional evidence that different tau PET spatial profiles might have more/less prognostic value in different stages of Alzheimer’s disease. This might also be relevant in the context of the age findings, where steeper MMSE decline was observed in older rather than younger participants. More in-depth analyses investigating the magnitude and spatial extent of tau PET in the context of age may have additional value when considering implementation in the clinical setting.

    Lastly, this study effectively demonstrates how a biomarker panel that includes amyloid, tau, and structural brain changes has utility for predicting future cognitive decline. It is important to acknowledge that this workup is costly, and that this cost is likely prohibitive for many patients. As the authors comment, to what extent these results will translate to blood plasma biomarkers is currently unknown and actively being investigated. As these results become available, it will be important to consider the trade-offs between affordability, access, and prediction accuracy for various biomarker modalities.

    References:

    Nelson PT, Alafuzoff I, Bigio EH, Bouras C, Braak H, Cairns NJ, Castellani RJ, Crain BJ, Davies P, Del Tredici K, Duyckaerts C, Frosch MP, Haroutunian V, Hof PR, Hulette CM, Hyman BT, Iwatsubo T, Jellinger KA, Jicha GA, Kövari E, Kukull WA, Leverenz JB, Love S, Mackenzie IR, Mann DM, Masliah E, McKee AC, Montine TJ, Morris JC, Schneider JA, Sonnen JA, Thal DR, Trojanowski JQ, Troncoso JC, Wisniewski T, Woltjer RL, Beach TG. Correlation of Alzheimer disease neuropathologic changes with cognitive status: a review of the literature. J Neuropathol Exp Neurol. 2012 May;71(5):362-81. PubMed.

    View all comments by Tobey Betthauser

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Tau PET Scans Turn Positive When Amyloid Does; Symptoms Follow
  2. Could Tau PET Replace Amyloid Biomarkers as a Diagnostic for AD?
  3. Flortaucipir Imaging Distinguishes Alzheimer’s From Other Disorders
  4. Forget Typical Alzheimer's: AI Finds Four Types.
  5. Where to Now, Phospho-Tau?
  6. Earliest of Them All: Blood P-Tau231 Assay Flags Pre-Amyloid Alzheimer’s
  7. Plasma P-Tau181 Predicts, Monitors Alzheimer’s Progression
  8. A Phospho-Tau Plasma Assay for Alzheimer’s?

Further Reading

Papers

  1. Devous MD Sr, Fleisher AS, Pontecorvo MJ, Lu M, Siderowf A, Navitsky M, Kennedy I, Southekal S, Harris TS, Mintun MA. Relationships Between Cognition and Neuropathological Tau in Alzheimer's Disease Assessed by 18F Flortaucipir PET. J Alzheimers Dis. 2021;80(3):1091-1104. PubMed.
  2. Buckley RF, Scott MR, Jacobs HI, Schultz AP, Properzi MJ, Amariglio RE, Hohman TJ, Mayblyum DV, Rubinstein ZB, Manning L, Hanseeuw BJ, Mormino EC, Rentz DM, Johnson KA, Sperling RA. Sex Mediates Relationships Between Regional Tau Pathology and Cognitive Decline. Ann Neurol. 2020 Nov;88(5):921-932. Epub 2020 Aug 31 PubMed.

Primary Papers

  1. Ossenkoppele R, Smith R, Mattsson-Carlgren N, Groot C, Leuzy A, Strandberg O, Palmqvist S, Olsson T, Jögi J, Stormrud E, Cho H, Ryu YH, Choi JY, Boxer AL, Gorno-Tempini ML, Miller BL, Soleimani-Meigooni D, Iaccarino L, La Joie R, Baker S, Borroni E, Klein G, Pontecorvo MJ, Devous MD Sr, Jagust WJ, Lyoo CH, Rabinovici GD, Hansson O. Accuracy of Tau Positron Emission Tomography as a Prognostic Marker in Preclinical and Prodromal Alzheimer Disease: A Head-to-Head Comparison Against Amyloid Positron Emission Tomography and Magnetic Resonance Imaging. JAMA Neurol. 2021 Aug 1;78(8):961-971. PubMed.

Annotate

To make an annotation you must Login or Register.