HAI Book 2025 - Flipbook - Page 197
Frontzkowski, Lukas
39
Developing a novel reference region for PI-2620-PET imaging to
facilitate assessment of 4-repeat tauopathies
Lukas Frontzkowski
1|2|3|4
Institute for Stroke and Dementia Research (ISD) | Department of Neurology, LMU University Hospital |
Department of Nuclear Medicine, LMU University Hospital | Department of Clinical Neurosciences and Cambridge
University Hospitals NHS Trust
Background: Neurodegenerative 4-repeat (4R) tauopathies commonly manifest as progressive supranuclear
palsy (PSP). PSP patients show elevated PI-2620-PET in subcortical 4R tau predilection sites (e.g., globus
pallidus), suggesting PI-2620-PET as a promising 4R tau neuroimaging candidate. However, optimal
quantification of PI-2620-PET in 4R tauopathies remains challenging, as conventional cerebellar tau-PET
reference regions also accumulate 4R tau. We aimed to use unbiased image-derived input function (IDIF) PET data
to determine an optimized PET reference region for in vivo quantification of 4R tau.
Methods: We obtained 60-minute dynamic PI-2620-PET in 54 PSP Richardson Syndrome (PSP-RS) patients and 19
healthy controls (HC), applying IDIF-modeling using carotid timeseries to assess unbiased PI-2620-PET binding
and determine total distribution volume (VT). Through an iterative approach, we intensity-normalized VT-images
against white-matter regions in the Hammers brain atlas, identifying regions where intensity-normalized pallidum
PET values showed the largest PSP-RS vs. HC differences. White-matter regions with strongest PSP-RS vs. HC
differences surviving multiple-comparison correction were summarized into a single reference region spanning
bilateral temporal white-matter. This ROI was then used to determine SUVRs using conventional 20-40 minute PI2620-PET data in PSP-RS, a PSP-non-RS validation sample (n=63), as well as non-tau disease controls (i.e., alphasynucleinopathies, n=20; Alzheimer9s disease, n=23).
Results: Using PI-2620 SUVRs obtained with the temporal-lobe white-matter reference, we detected strong PSPRS vs. HC group differences in basal ganglia SUVRs using voxel-wise comparisons (p
