Researchers at Japan's National Institutes for Quantum Science and Technology (QST) report that in progressive supranuclear palsy (PSP), frontal thinking problems track the connectivity between each person's tau deposits and a shared cortical circuit, rather than the local amount of tau. The study, led by Yuki Hori and Toshiyuki Hirabayashi, was published in Science Advances on 10 July 2026, after a preprint version was posted on bioRxiv in February 2025 and revised in October 2025.
The team scanned 37 patients with clinically diagnosed PSP (22 men and 15 women, mean age 70 years, mean 3.4 years since symptom onset, mean PSP rating scale score 41) using 18F-florzolotau, a tau PET tracer developed at QST, together with structural MRI. All 37 showed tau signal and none showed amyloid-beta on Pittsburgh compound-B scanning. Because the patients themselves were not scanned during rest for connectivity, the researchers used each person's tau locations as seeds in a public resting-state fMRI dataset from 100 healthy volunteers, the Brain Genomics Superstruct Project, to estimate which distant regions those deposits connect to.
Although the tau maps differed from patient to patient and sat mainly in subcortical structures, the seeds converged on a common cortical set that the authors call the PSP-tau network: lateral prefrontal cortex, dorsal anterior cingulate cortex, anterior insula and posterior parietal cortex. That network could not be reproduced from atrophy locations alone. Across its nodes, more than 80 percent of areas correlated most strongly with Frontal Assessment Battery based cognitive scores, while local subcortical tau burden, in the globus pallidus and midbrain, correlated with motor scores instead. Hirabayashi said the group set out to test "whether local tau pathology could affect distant brain regions through neural circuits".
The design carries clear limits. This is a single cross-sectional observation in a small group at one research centre, the connectivity is inferred from other people's healthy brains rather than measured in the patients, and the findings are correlations, not demonstrated cause. The tracer is one of several used for four-repeat tauopathies, and results with a different probe might not line up exactly.