The phosphorylation of tau in the Alzheimer’s brain is not a binary state. It evolves over time, passing through distinct stages that reflect worsening disease, according to a study in the March 11 Nature Medicine led by Randall Bateman and Eric McDade at Washington University School of Medicine in St. Louis.

  • CSF p-tau217 and p-tau181 indicate presence of amyloid plaques in brain.
  • CSF p-tau205 correlates with atrophy and sluggish brain metabolism later in AD.
  • Tangles arrive even later, correlating with cognitive decline.

The scientists analyzed phospho-tau isoforms and total tau in the cerebrospinal fluid (CSF) of people who carried familial AD mutations. As their disease progressed, different p-tau forms became detectable that correlated with imaging markers of metabolism, atrophy, and amyloid. The researchers identified three stages. In the first, soluble p-tau217 and p-tau181 rose in CSF in tandem with amyloidosis in the brain. (Yes, p-tau and plaques rose simultaneously, not p-tau and tangles.) In the second stage, rising CSF p-tau205 correlated with waning brain metabolism and shrinking gray matter. In the third stage, CSF total tau levels spiked; at this point, tangles formed and cognitive decline began.

“These data have implications for clinical trials, because tau therapies might have different impacts depending on the stage of AD,” Bateman told Alzforum. He believes these soluble tau biomarkers could be used to stage disease in individual patients, and perhaps to predict their progression. First author Nicolas Barthélemy presented some of these findings at the Alzheimer’s Association International Conference in Los Angeles (Aug 2019 conference news). 

Other researchers saw the implications as well. Gil Rabinovici at the University of California, San Francisco, said the data strengthen the idea that phosphorylated tau accumulates in response to Aβ buildup in the brain. Henrik Zetterberg at the University of Gothenburg, Sweden, agreed. He noted that the study adds knowledge by identifying specific p-tau isoforms that act as early amyloid response markers.

Three Stages of Tau? Changes in CSF tau suggest distinct responses of the molecule in different phases of AD pathogenesis: amyloidosis, neuronal dysfunction, and neurodegeneration. [Courtesy of Barthélemy et al., Nature Medicine.]

Previously, researchers distinguished between two main forms of soluble tau: p-tau and total tau. In the amyloid/neurofibrillary tangle/neurodegeneration, aka ATN, system for staging AD, p-tau was grouped together with tau PET as a marker of neurofibrillary tangles, and total tau with atrophy as an indicator of neurodegeneration (Aug 2016 news). However, recent data have raised doubt about this scheme. For example, some studies of AD patients found little relationship between CSF p-tau and tangles (Mattsson et al., 2017; La Joie et al., 2018). 

To dissect how CSF tau relates to other AD pathologies, Barthélemy analyzed longitudinal CSF samples from 370 participants in the Dominantly Inherited Alzheimer Network (DIAN). In this group, 152 people carried a familial AD mutation but were asymptomatic, 77 people were symptomatic carriers, and 141 were noncarriers. The researchers measured several specific p-tau isoforms using mass spectrometry.

Among carriers, CSF p-tau217 rose first, at 21 years before the estimated year of onset (EYO). Next came p-tau181, at 19 years before. Both paralleled the progression of amyloidosis. This occurs decades before tau PET scans turn positive. “Amyloid and these p-tau forms go up at the same time in the brain,” Bateman said. Of the two, p-tau217 had the closest relationship with plaques, correlating most strongly with both global and regional accumulation as measured by amyloid PET. In fact, the authors found that p-tau217 was a more sensitive indicator of brain amyloidosis than were commercial Elecsys tests for p-tau181 or total tau.

The picture was quite different for CSF p-tau205. It began to climb 13 years before a person’s EYO. This marker correlated with waning brain metabolism as seen by FDG PET. It also closely linked to atrophy measured by structural MRI. Atrophy associated with p-tau217 and total tau as well, but less strongly. Total tau began to creep up 17 years before EYO. P-tau205 and total tau may together mark the second stage of AD, when synaptic loss and degeneration begin, Bateman suggested.

The third stage was defined by a positive tau PET signal, indicating the presence of widespread tangles. In DIAN participants, and as measured with flortaucipir, this occurs right around the EYO. CSF total tau correlated with the tau PET signal in limbic and posterior cortex, as did p-tau205. Conversely, CSF p-tau217 and p-tau181 dropped as tangles accumulated.

Overall, the findings belie the idea that CSF phospho-tau represents tangles and CSF total tau represents neuronal death. “This paper challenges previous assumptions about phospho-tau and total tau,” Rabinovici said. “We need to revisit some of those assumptions and refine the ATN framework.”

To see if the findings held in sporadic AD, the authors analyzed cross-sectional data from 102 people seen at WashU: 39 were amyloid-negative, 18 amyloid positive but asymptomatic, and 45 were amyloid positive and symptomatic. They found a similar pattern in the CSF: p-tau181 and p-tau217 were high whenever amyloidosis was present, while p-tau205 and total tau marked later stages of disease. Compared with DIAN participants, levels of p-tau and total tau were lower, however. “Even though some of the analyses were corroborated, it’s still important to be careful about generalizing results from the DIAN population to the more common sporadic disease,” Rabinovici noted.

What do these data reveal about disease processes? One major implication is that amyloidosis directly triggers tau phosphorylation, without any time lag, Bateman said. He thinks this represents a physiological, adaptive response to amyloid plaques. Plaques crowd the intercellular space, damaging axons, he noted. This may provoke kinases to phosphorylate tau, causing the molecule to detach from microtubules. Microtubules then break down, causing damaged axons to retract. Later, neurons may extend axons in new directions, bypassing plaques. In support of this, Bateman and colleagues previously found that amyloid plaques spur neurons in the AD brain to double total tau production (Mar 2018 news). Similarly, tau production increases in animal models of amyloidosis in the absence of any neurofibrillary tangles (Maia et al., 2013). 

Eventually, however, this adaptive response fails, and tau builds up in tangles. This may be why p-tau181 and p-tau217 begin to dwindle as tangles form, Bateman suggested.

Zetterberg agrees with this model. “The amazing resistance [to amyloid], in terms of lack of symptoms in individuals when they first become amyloid-positive, [means] the brain is trying to compensate,” Zetterberg said.

The model explains why CSF p-tau does not rise in other tauopathies (Hampel et al., 2004; Hu et al., 2013). “In some sense, CSF phospho-tau is an amyloid marker,” Zetterberg said. This suggests that an anti-amyloid treatment that lowers Aβ toxicity would suppress p-tau217 and p-tau181, he added. Thus, these markers could provide a readout of drug efficacy. “Tau is now more complicated as a biomarker, but more interesting as a theragnostic,” Zetterberg said.

Questions remain, such as how well CSF p-tau isoforms reflect tau pathology in the brain, and what effect anti-tau therapy would have on these soluble isoforms. To answer the first question, Bateman and colleagues are analyzing phosphorylated tau species in postmortem brain tissue and correlating with levels in the CSF. To answer the second, they will measure changes in CSF tau in the anti-amyloid and future anti-tau arms of the DIAN-TU trial. It is unclear if tau therapy will affect these isoforms, and if it does, whether that would be good or bad.

“It will be important to clarify if the biomarker changes in soluble tau are on a causal pathway that leads to development of aggregated tau,” noted Niklas Mattsson at Lund University, Sweden. If so, it would suggest anti-tau therapy could be protective at this early stage of disease.

At last year’s AAIC, Mattsson presented similar data to that from the WashU group, linking phospho-tau to amyloid but not tangles. “We also have new results from patients with sporadic AD, patients with MAPT mutations, and transgenic AD mice, that are strongly convergent with the findings by Barthélemy et al.,” Mattsson wrote to Alzforum. Zetterberg suggested that knowing specific tau phosphorylation sites are involved in the responses to amyloid will allow researchers to develop inhibitors against p-tau181 and p-tau217, if these forms of tau are shown to be harmful.

Rabinovici, meanwhile, is intrigued by the potential of these phospho-tau epitopes to aid diagnosis. “It might be useful, instead of looking at one phosphorylated epitope at a time, to think about panels of phospho-tau forms that might give us more information about disease stage and prognosis in individual patients,” he suggested.—Madolyn Bowman Rogers


  1. The finding that early changes in soluble tau, including increased phosphorylation, start in parallel with Aβ aggregation is very interesting. This shows how closely integrated the key hallmarks of AD are, even in the earliest stage of the disease. It is intriguing that these events appear to precede widespread development of aggregated tau (and corresponding atrophy and cognitive impairment) by several years.

    Further studies are now needed to understand the cellular mechanisms that may link the early changes in aggregated Aβ with the changes in soluble tau, and with downstream events in AD. For example, it will be important to clarify if the biomarker changes in soluble tau are on a causal pathway that leads to development of aggregated tau. If we can establish such links, it would argue for anti-tau interventions in individuals with CSF signs of soluble tau alterations, but without pathological tau PET. Alternatively, the changes in soluble tau may represent a biochemical epiphenomenon that may be very useful for diagnostic purposes (perhaps especially when measured in plasma or serum) but without major importance for disease progression.

    Overall, I think the findings are very much in line with previous findings from us and others on the associations between CSF p-tau with tau PET, which have suggested that changes in soluble tau may occur before significant tau PET changes (e.g. Mattsson et al., 2017; Mattsson et al., 2018). We also have new results from patients with sporadic AD, patients with MAPT-mutations, and from transgenic AD-mice, that are strongly convergent with the findings by Barthélemy et al. We describe these results in a paper currently in press in Science Advances, scheduled for publication in April. 


    . 18F-AV-1451 and CSF T-tau and P-tau as biomarkers in Alzheimer's disease. EMBO Mol Med. 2017 Sep;9(9):1212-1223. PubMed.

    . Comparing 18 F-AV-1451 with CSF t-tau and p-tau for diagnosis of Alzheimer disease. Neurology. 2018 Jan 30;90(5):e388-e395. Epub 2018 Jan 10 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Move Over Aβ, CSF P-Tau Tells Us There’s Plaque in the Brain
  2. Staging of Alzheimer’s, the Second: Neurodegeneration Does Not Equal Tauopathy
  3. Isotope Labeling Links Tau Production to Aβ Burden

Paper Citations

  1. . 18F-AV-1451 and CSF T-tau and P-tau as biomarkers in Alzheimer's disease. EMBO Mol Med. 2017 Sep;9(9):1212-1223. PubMed.
  2. . Associations between [18 F]AV1451 tau PET and CSF measures of tau pathology in a clinical sample. Neurology. 2018 Jan 23;90(4):e282-e290. Epub 2017 Dec 27 PubMed.
  3. . Changes in amyloid-β and Tau in the cerebrospinal fluid of transgenic mice overexpressing amyloid precursor protein. Sci Transl Med. 2013 Jul 17;5(194):194re2. PubMed.
  4. . Measurement of phosphorylated tau epitopes in the differential diagnosis of Alzheimer disease: a comparative cerebrospinal fluid study. Arch Gen Psychiatry. 2004 Jan;61(1):95-102. PubMed.
  5. . Reduced CSF p-Tau181 to Tau ratio is a biomarker for FTLD-TDP. Neurology. 2013 Nov 26;81(22):1945-52. Epub 2013 Oct 30 PubMed.

Further Reading

Primary Papers

  1. . A soluble phosphorylated tau signature links tau, amyloid and the evolution of stages of dominantly inherited Alzheimer's disease. Nat Med. 2020 Mar;26(3):398-407. Epub 2020 Mar 11 PubMed.