The hunt for PET tracers that can detect four-repeat tau deposits has turned up two ligands, PI-2620 and APN-1607, that bind these fibrils, but neither has yet been validated for diagnostic use. In a May 7 preprint on bioRxiv, researchers led by Matthias Brendel, Nicolai Franzmeier, Sigrun Roeber, and Günter Höglinger of Ludwig Maximilians University of Munich strengthen the case that Life Molecular Imaging’s PI-2620 detects 4R tau in living brains. This form of tau accumulates in tauopathies such as progressive supranuclear palsy and corticobasal degeneration. The authors report that in six PSP patients, the PET signal during life correlated with 4R tau deposits in their brains detected by autoradiography and immunostaining after death. The signal came from neurons and oligodendrocytes, but not astrocytes. The authors identified the gray/white-matter boundary in the frontal cortex as having the strongest PET signal, suggesting that focusing on this region could sharpen PI-2620’s ability to discriminate 4R tauopathies from healthy controls.

  • PI-2620 PET signal correlates with 4R tau deposits in postmortem brain.
  • The PET signal arises from aggregates in neurons and oligodendrocytes.
  • The gray/white matter boundary in frontal cortex gives the strongest signal.

While calling the findings an advance, scientists remained agnostic about PI-2620’s diagnostic utility for 4R tauopathies. “This elegant study … shows a strong correlation of regional PI-2620 signals to abundance of fibrillary tau,” Neil Vasdev at the University of Toronto wrote to Alzforum. Rik Ossenkoppele at Vrije University Amsterdam agreed the study was well-done, but cautioned there remains substantial overlap between the PET signal in 4R tauopathy patients and controls. “I am not convinced yet that the in-vivo signal of this tracer is strong and robust enough for offering diagnostic and/or prognostic value at the individual level in suspected primary 4R tauopathies,” he wrote (comments below).

Besides PI-2620 and APN-1607, other 4R tau tracers are in development. Vasdev and colleagues Chet Mathis at the University of Pittsburgh and Samuel Svensson at biotech Oxiant Discovery in Södertälje, Sweden, have identified a candidate, OXD-2314.

Lit Up: 4R Tau. Correlation of the PET signal (bottom, red) in basal ganglia of six PSP patients with postmortem AT8 staining (top) and PI-2620 autoradiography (middle) of tissue from those regions. [Courtesy of Slemann et al., bioRxiv.]

Previously, Brendel and colleagues reported that PI-2620 lit up basal ganglia and frontal cortex in PSP patients, and bound to those sections in banked brain samples. The tracer distinguished typical PSP patients from controls with 85 percent sensitivity and 77 percent specificity, but the PET signal was weak (Jul 2020 news). Other studies reported mixed findings for whether the tracer recognized 4R tau in postmortem brain (Malarte et al., 2023; Aguero et al., 2024). 

To gather more comprehensive data, joint first authors Luna Slemann, Johannes Gnörich, Selina Hummel, and Laura Bartos first used 18FPI-2620 to scan PS19 mice, which carry P301S mutant human tau and accumulate 4R aggregates starting at about 6 months of age. By 10 months, the PET signal in PS19 mice was distinguishable from that in control mice, being 20 percent higher. To find out what cell types contributed to the signal, the authors isolated neurons and astrocytes immediately after administering the PET tracer, and measured the radioactivity in each. All the signal came from neurons, with PS19 neurons having about twice as much tracer binding as those from wild-type mice.

Moving to people, Slemann and colleagues found that in the six PSP patients who had been scanned during life and donated their brains, the PET signal correlated with autoradiography of tissue isolated from their basal ganglia, and with tau aggregates as judged by AT8 staining. PET scans had been done an average of two years before death; tighter timing might have improved the radiography and tangle correlations of 0.89 and 0.82, respectively, the authors noted.

The authors zeroed in on specific cell types in human tissue, comparing AT8 staining with PI-2620 autoradiography in frontal cortex sections from 16 additional PSP brains. There were more tau deposits in astrocytes than in neurons and oligodendrocytes as per immunohistochemistry; even so, the PET signal came exclusively from the latter two. Likely this is because neuronal and oligodendrocyte deposits are denser, boosting the PET signal, Franzmeier told Alzforum. Astrocytes' many fine processes scatter their tau deposits over a broad area, rendering the PET signal too faint to see. Tangle-bearing “tufted” astrocytes are a hallmark of PSP.

In line with this cellular distribution, the authors found that the PET signal was strongest near the gray/white-matter boundary in the frontal cortex, probably because of the density of oligodendrocytes in this region. The signal there distinguished 17 PSP scans from nine control scans with a Cohen’s d effect size of 1.68, compared with 1.37 when using the gray-matter signal. Ideally, a diagnostic scan would combine PET with MRI to quantify the signal in the gray/white-matter boundary, Franzmeier said.

Makoto Higuchi at the National Institute of Radiological Sciences, Japan, noted that technical issues, such as radioactivity spillover from the extracranial space, still dog PI-2620 imaging. These would need to be addressed before assessing the tracer’s reliability, he wrote. Higuchi also wanted to know how the gray/white-matter boundary performs diagnostically for discriminating PSP cases and controls (comment below). Franzmeier said they are now assessing the clinical utility of PI-2620 visual reads.

Meanwhile, Vasdev, Mathis, and Svensson previously developed a PET tracer, OXD-2115, with high affinity for 4R tau (Lindberg et al., 2021). Because that tracer poorly entered brain, they developed analogs with better uptake. Their current lead candidate, OXD-2314, performed well in rats and non-human primates, and will now start human trials, Vasdev told Alzforum.—Madolyn Bowman Rogers


  1. 18FPI-2620 was developed to image aggregated tau with mixed 3- and 4-microtuble-binding repeat domains (3R- and 4R-tau) in patients with Alzheimer’s disease (AD), and has recently been applied to patients with 4R-tauopathies, such as progressive supranuclear palsy (PSP). This elegant study sheds light on the mechanism of binding of 18FPI-2620 with PET and autopsy studies that show a strong correlation of regional 18FPI-2620 signals to abundance of fibrillary tau. Furthermore, in vivo imaging with 18FPI-2620 in a transgenic mouse model of 4R-tau was followed by an innovative cell-sorting approach to determine radiotracer uptake by cell type (neurons vs. astrocytes) and validates the claims of high specificity of this radiopharmaceutical for intraneuronal tau in human PET studies. It will be interesting to apply the cell-sorting method to next-generation tau-PET radiotracers that our laboratories and others are developing for non-AD tauopathies, such as 18FOXD-2314, which is planned for first-in-human testing at our center in the next few weeks.

  2. Like other tau PET tracers, for example MK6240, flortaucipir, and RO948, PI2620 has shown excellent diagnostic performance in distinguishing symptomatic AD from other neurodegenerative disorders. An open question is whether PI2620 can also be utilized for diagnostic purposes in individuals with a primary 4R tauopathy, such as progressive supranuclear palsy (PSP).

    This preprint describes very elegant and well-executed experiments showing 1) increasing PI2620 signal in a 4R (PS19) mouse model in the presence of high neuronal tau, 2) good correlations between antemortem PI2620 signal versus postmortem fibrillary tau in autopsy samples of PSP patients, and 3) that the PI2620 signal in PSP is mainly driven by tau-positive neurons and oligodendrocytes (but not astrocytes) as defined by autoradiography.

    This main implication of this work is that it provides a better understanding about the source of PI2620 binding in individuals with PSP.  I am not convinced yet that the in vivo signal of this tracer is strong and robust enough for offering diagnostic and/or prognostic value at the individual level in suspected primary 4R tauopathies. This is based on the substantial signal overlap described in previous studies between PSP cases and controls/other movement disorders, even in PSP signature regions such as the globus pallidus and putamen. However, I am happy to be proven wrong, since robust molecular biomarkers for primary tauopathies are desperately needed.

  3. In this work, Slemann and colleagues utilized tau-transgenic mice and brain sections derived from PSP patients who had undergone PET scans, along with sections from a brain bank, to demonstrate the ability of a tau PET probe, 18FPI2620, to capture four-repeat (4R) tau aggregates in the brain. The non-clinical assessments in animals would provide unequivocal evidence for the reactivity of the probe with the target components, since in vivo PET and postmortem data could be compared in the same subject with a minimal chronological gap between the two assays. While the authors’ efforts to prove the binding of 18FPI2620 to 4R tau assemblies are acknowledged, several methodological issues need to be figured out before reaching any conclusions.

    The PS19 transgenic mice employed in the PET analysis develop only a limited number of densely packed, thioflavin-S- and FSB-positive neuronal tau inclusions in the hippocampus, entorhinal, and retrosplenial cortices, and in the amygdala. These may not be readily detectable by PET, despite numerous AT8-positive phosphorylated tau deposits in these locations. Such mature tau deposits are more abundant in the brainstem and spinal cord, but PET of these pathologies is often challenging due to small volumes of the target anatomical structures. In the current investigation, marked radioactivity spillover from neighboring extracranial space was noted in the cerebellum, brainstem, amygdala, entorhinal cortex, and olfactory bulb of transgenic and wild-type mice, although the amount of radio signal outside the brain was not shown in the trimmed PET images. The spillover effect could be more pronounced if the brain parenchyma displays atrophy. Even with the authors taking great care in defining the regions of interest by eroding them, the radioactivity spread from surrounding tissues may be nearly unavoidable.

    The measurement of radioactivity uptake in isolated neurons and astrocytes are also interesting approaches, however, the normalization of the observed values may not be simple. The radiotracer uptake per single cell was determined as percent injected dose (ID)/g sample, which might be influenced by the body weight of the animal (increased by the weight loss) and by the number and volumes of the cells in the unit tissue (possibly decreased in astrocytes if they are hypertrophic and proliferate). Accordingly, the percent ID/g values in the cellular assay may not be directly compared with the target-to-reference ratio of the radioactivity retention (SUVR) in PET quantifications.

    The autoradiography of the samples from scanned subjects offers highly valuable information on the molecular and cellular sources of the in vivo radio signals, and the establishment of robust techniques for evaluating the tracer binding in the brain sections should be an essential requirement. It remains to be assessed whether the autoradiographic labeling of subregions enriched with tau pathologies can be displaced by an excess concentration of non-radiolabeled PI-2620. It should also be clarified why basal ganglia and frontal cortex sections equally exhibited low radiotracer binding in the control brain despite profound, nonspecific, in vivo radioactivity retention in the basal ganglia and thalamus. Of note, the intensification of the autoradiographic signals around the boundary between gray and white matter of the frontal cortex was in line with PET data showing the locally highest Cohen’s d (0.5 – 0.7) in this cortical segment for the separation between PSP patients and controls. One would like to know the diagnostic significance of this finding, as this effect size usually translates to an AUC of ~0.7.

    Studies have shown that tau aggregates in PSP and other 4R tauopathies and 4R tau-expressing tau transgenics, e.g., rTg4510 mice, can be visualized with high contrast by PET with 18Fflorzolotau, aka APN-1607/PM-PBB3, which binds to the cross-β structure commonly shared by diverse tau fibril folds. Besides this radiotracer, PET probes reacting with fibrils in primary tauopathies that have a different binding mode would exert superb diagnostic performance with high selectivity for tau versus other protein assemblies. Evaluations of 18FPI-2620 may bring clues to the development of such tau ligands, and refinements of PET and autoradiographic examinations in the research community will be a high-priority issue.

Make a Comment

To make a comment you must login or register.


News Citations

  1. At Holloway Summit, FTD Imaging Shows New Vista
  2. PET Tracer PI-2620 Detects 4R Tau Deposits

Research Models Citations

  1. Tau P301S (Line PS19)

Paper Citations

  1. . Discriminative binding of tau PET tracers PI2620, MK6240 and RO948 in Alzheimer's disease, corticobasal degeneration and progressive supranuclear palsy brains. Mol Psychiatry. 2023 Mar;28(3):1272-1283. Epub 2022 Nov 29 PubMed. Correction.
  2. . Head-to-head comparison of [18F]-Flortaucipir, [18F]-MK-6240 and [18F]-PI-2620 postmortem binding across the spectrum of neurodegenerative diseases. Acta Neuropathol. 2024 Jan 27;147(1):25. PubMed.
  3. . Radiosynthesis, In Vitro and In Vivo Evaluation of [18F]CBD-2115 as a First-in-Class Radiotracer for Imaging 4R-Tauopathies. ACS Chem Neurosci. 2021 Feb 17;12(4):596-602. Epub 2021 Jan 26 PubMed.

Further Reading

Primary Papers

  1. . Neuronal and oligodendroglial but not astroglial tau translates to in vivo tau-PET signals in primary tauopathies. 2024 May 07 10.1101/2024.05.04.592508 (version 1) bioRxiv.