Summary: While tau PET scans are vital for “seeing” Alzheimer’s biology in living patients, they have a known quirk: the most common tracer, Flortaucipir (Tauvid), often “lights up” even when tau tangles aren’t the primary driver. In a high-precision study, researchers used AI-enabled, voxel-by-voxel alignment to compare PET scans directly with postmortem brain tissue.
They discovered that in non-Alzheimer’s cases and even some controls, the PET signal was actually reacting to iron deposits and MAO-B (a marker of neuroinflammation) rather than tau itself. This “off-target” binding explains why scans can sometimes over-predict the presence of toxic tangles.
Key Facts
- The AI Matchmaker: Researchers used AI to computationally match thousands of points on a PET scan to the exact corresponding tissue location under a microscope, providing unprecedented anatomical precision.
- The “False Positive” Trio: In non-Alzheimer’s disorders, the PET signal was better explained by ferric iron and MAO-B (linked to reactive astrocytes) than by phospho-tau.
- Alzheimer’s vs. Others: In Alzheimer’s disease, tau remains a major contributor to the signal, but it is not the only one. In other tauopathies, the tracer struggles to differentiate between tau and inflammation.
- Refining the “Glow”: The findings help clinicians avoid “over-interpretation” of borderline signals, which is critical for prognosis and selecting patients for drug trials.
- Path to Next-Gen Tracers: By identifying these specific “off-target” culprits, the study informs the development of future tracers that will be 100% specific to tau.
Source: UCSF
Tau proteins play an important role in Alzheimer’s disease. Tau helps to stabilize neurons in the brain, but in Alzheimer’s disease, tau proteins can misfold and tangle inside neurons. These tangles spread across the brain forming toxic clumps that impair neuronal function and ultimately lead to cell death.
PET scans are one of the few ways that clinicians and researchers can “see” Alzheimer’s disease biology in a living person. PET scans for tau are being increasingly used in specialty clinics and research to estimate the burden and distribution of tau tangles, providing information that can influence diagnostic confidence, prognosis discussions, and how patients are selected for novel Alzheimer’s therapies and tracked in Alzheimer’s drug trials.
At the same time, there is an important limitation to tau PET scans: the most widely used tau PET tracer (Flortaucipir/Tauvid) can also bind to other biological features in the brain, sometimes showing a low-level signal on scans even when tau is not the main driver of the results. This is especially true in non-Alzheimer’s disorders, where tau protein accumulations show a different structure than in Alzheimer’s disease.
In a study appearing February 25 in Acta Neuropathologica, UC San Francisco researchers use a rare combination of patient tau PET imaging plus detailed postmortem brain tissue mapping, to clarify the extent to which non-tau factors influence the Flortaucipir tracer signal.
With the aid of AI-enabled, voxel-by-voxel alignment between the PET scan and microscopy, the researchers computationally matched the PET signal to the corresponding tissue location and generated thousands of point-by-point comparisons per person, giving much higher anatomical precision than typical PET–autopsy studies.
They then measured, side-by-side, the tissue features most relevant to interpreting flortaucipir: phospho-tau (tau pathology), ferric iron deposits and MAO-B, a marker linked to reactive astrocytes which are a key component of neuroinflammation. They found that in several non-Alzheimer’s tauopathies (and in a non-tau control case), the PET signal was better explained by iron and/or MAO-B–related processes than by tau itself.
“Our results help explain why tau PET scans sometimes ‘light up’ beyond what tau pathology alone would predict,” says co-senior author Lea T. Grinberg, MD, PhD, a former UCSF neuroscientist and current professor of Neuroscience at the Mayo Clinic .“In Alzheimer’s disease, tau is a major contributor to flortaucipir signal—but not the only one.”
The researchers believe these findings can help clinicians avoid over-interpretation of borderline signal in a tau PET scan and improve how scans are used for prognosis and trial decisions.
“This work doesn’t argue against tau PET, rather, it helps clinicians and researchers interpret it more accurately,” said co-senior author Gil Rabinovici, MD, the Edward Fein and Pearl Landrith Distinguished Professorship in Memory & Aging in the UCSF Department of Neurology.
“By clarifying when and where the signal reflects tau versus other biology, the findings can support better clinical decision-making and can inform the development of next-generation tracers that are more specific.”
Additional Authors: Yuheng Chen, Renaud La Joie, Felipe L. Pereira, Ganna Blazhenets, Lucile Zhu, Salvatore Spina, William W. Seeley, Helmut Heinsen, Daniela Ushizima, Duygu Tosun
Funding: The work was supported by a research grant from Eli Lilly (Lilly Research Award Program) and NIA R01AG070826, NIA K24 AG053435 (Grinberg), P30 AG062422 (GDR), U01 AG057195, P01AG09724, and the Rainwater Charitable Foundation (GDR).
Key Questions Answered:
A: Not necessarily. If you have Alzheimer’s, the signal is still largely driven by tau. However, this study gives doctors a “decoder ring” to understand when a low-level glow might actually be inflammation or iron rather than the disease spreading.
A: Flortaucipir is designed to find the specific shape of tau tangles. Unfortunately, the chemical structure of iron deposits and the MAO-B enzyme can “trick” the tracer into binding to them as well, causing those areas to light up on the scan.
A: Actually, it’s great news. This is about “sharpening the tool.” By knowing exactly what else the scanner is seeing, researchers can develop more accurate software to filter out the noise and design next-generation tracers that only look for tau.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this neuroimaging and Alzheimer’s disease research news
Author: Melinda Krigel
Source: UCSF
Contact: Melinda Krigel – UCSF
Image: The image is credited to Neuroscience News
Original Research: Open access.
“Disentangling on and off-target binding in flortaucipir PET: a voxel-to-voxel P-tau, ferric iron, and MAO-B histology-to-flortaucipir PET comparison” by Yuheng Chen, Renaud La Joie, Felipe L. Pereira, Ganna Blazhenets, Lucile Zhu, Salvatore Spina, William W. Seeley, Helmut Heinsen, Daniela Ushizima, Duygu Tosun, Lea T. Grinberg, and Gil Rabinovici. Acta Neuropathologica
DOI:10.1007/s00401-026-02983-x
Abstract
Disentangling on and off-target binding in flortaucipir PET: a voxel-to-voxel P-tau, ferric iron, and MAO-B histology-to-flortaucipir PET comparison
Flortaucipir PET imaging has significantly advanced our ability to visualize tau pathology in vivo. However, off-target Flortaucipir signal remains a considerable challenge for interpreting of imaging results, particularly in non-Alzheimer’s tauopathies and non-tau pathologies.
To better understand this off-target signal, we used an innovative voxel-to-voxel correlation approach, analyzing thousands of histology-Flortaucipir pairs from individual cases. This allowed us to quantitatively assess the relationship between Flortaucipir PET signal and three key biological factors: histological tau burden (CP-13 phospho-tau), ferric iron (Perls’ Prussian blue), and monoamine oxidase B (MAO-B).
Our study included individuals with Alzheimer’s disease (AD), various non-AD tauopathies, and a case of FTLD-TDP-43 type A. In AD, Flortaucipir signal showed a significant but moderate correlation with histological tau pathology, suggesting that while tau is a major contributor, other biological factors also influence Flortaucipir binding in AD.
Conversely, in non-AD tauopathies and FTLD-TDP-43, correlations between Flortaucipir signal and tau pathology were weak or negligible. Instead, Flortaucipir signal correlated more strongly with ferric iron and MAO-B.
However, these factors did not fully explain all the off-target signals, implying other unknown contributors are likely involved. These findings underscore the complexity of interpreting Flortaucipir PET scans.
A thorough understanding of off-target binding mechanisms is crucial for improving the diagnostic accuracy of Flortaucipir PET and its specificity.
