In vivo PET classification of tau pathologies in patients with frontotemporal dementia

Abstract Frontotemporal dementia refers to a group of neurodegenerative disorders with diverse clinical and neuropathological features. In vivo neuropathological assessments of frontotemporal dementia at an individual level have hitherto not been successful. In this study, we aim to classify patients with frontotemporal dementia based on topologies of tau protein aggregates captured by PET with 18F-florzolotau (aka 18F-APN-1607 and 18F-PM-PBB3), which allows high-contrast imaging of diverse tau fibrils in Alzheimer’s disease as well as in non–Alzheimer’s disease tauopathies. Twenty-six patients with frontotemporal dementia, 15 with behavioural variant frontotemporal dementia and 11 with other frontotemporal dementia phenotypes, and 20 age- and sex-matched healthy controls were included in this study. They underwent PET imaging of amyloid and tau depositions with 11C-PiB and 18F-florzolotau, respectively. By combining visual and quantitative analyses of PET images, the patients with behavioural variant frontotemporal dementia were classified into the following subgroups: (i) predominant tau accumulations in frontotemporal and frontolimbic cortices resembling three-repeat tauopathies (n = 3), (ii) predominant tau accumulations in posterior cortical and subcortical structures indicative of four-repeat tauopathies (n = 4); (iii) amyloid and tau accumulations consistent with Alzheimer’s disease (n = 4); and (iv) no overt amyloid and tau pathologies (n = 4). Despite these distinctions, clinical symptoms and localizations of brain atrophy did not significantly differ among the identified behavioural variant frontotemporal dementia subgroups. The patients with other frontotemporal dementia phenotypes were also classified into similar subgroups. The results suggest that PET with 18F-florzolotau potentially allows the classification of each individual with frontotemporal dementia on a neuropathological basis, which might not be possible by symptomatic and volumetric assessments.

For PET scans with Biograph mCT, the mCT flow system provided 109 sections with an axial field-of-view (FOV) of 21.8 cm.The intrinsic spatial resolution was 5.9 mm in-plane and 5.5 mm full-width at half-maximum axially.Images were reconstructed using a filtered back projection algorithm with a Hanning filter (4.0 mm full-width at half-maximum).Attenuation correction was applied based on computed tomography images, and correction for randoms was performed using the late coincidence counting method.
For PET scans with an ECAT Exact HR+, the HR+ system provided 63 sections with an axial FOV of 15.5 cm, and contiguous 2.46-mm slices with 5.6-mm transaxial and 5.4-mm axial resolution.Images were reconstructed by a filtered back-projection method (Hanning filter, cutoff frequency: 0.4 cycle/pixel).Attenuation correction was applied for each image using 10-min transmission scan data with a 68 Ge-68 Ga line source.
2) 18 F-florzolotau PET 90 min after an intravenous rapid bolus injection of 18 F-florzolotau in a dim room to avoid photo racemization, a 20-min PET acquisition (4×5-min frames) was performed with the Biograph mCT flow system (see 11 C-PiB PET section for scan parameters).
We also performed an exploratory voxel-based analysis of 18 F-florzolotau SUVR images to identify brain regions with enhanced radiotracer retentions in either threerepeat tau isoform (3RT)-like or four-repeat tau isoform (4RT)-like behavioral variant FTD (bvFTD) patients compared to the controls.Two sample t-tests were performed in SPM12, and the same statistical thresholds and the cortical mask, as mentioned above, were used.

2) Patients with other FTD phenotypes
In patients with other FTD phenotypes, one PPA patient (PPA1), three of four PNFA patients (PNFA1, PNFA2, PNFA4), and all CBS patients (CBS1-CBS3) showed regions with significantly higher SUVR values as compared to the controls, with both voxel-level height threshold of p < 0.001, uncorrected (> 20 voxels), and cluster-level threshold of pFWE < 0.05.Other patients (PPA2, PPA3, PNFA3, and PSP1) did not present cortical regions with significantly higher SUVR values with either of the statistical thresholds compared to the controls.
As shown in Supplementary Figure 2, patient PPA1 demonstrated significantly higher SUVR in relatively widespread regions in the inferior frontal and inferior and middle temporal cortices.In three of four PNFA patients (PNFA1, PNFA2, and PNFA4), significantly higher SUVR values were observed in the precentral cortex and/or parietal GM and WM regions.The regions with higher SUVR values in patient PNFA1 exhibited left-hemisphere dominance, while those in patient PNFA2 showed right-hemisphere dominance.On the other hand, no clear hemispheric dominance was found in patient PNFA4.Two Aβ-negative CBS patients (CBS1 and CBS2) demonstrated significantly higher SUVR values in the precentral cortex and surrounding WM regions, with left-hemisphere dominance.The Aβ-positive patient with CBS (CBS3) showed significantly higher SUVR values in widespread cortical regions.
3) Subgroups of 3RT-like and 4RT-like bvFTD patients as compared to controls As shown in Supplementary Figure 3, 3RT-like patients with bvFTD (P01-P03) showed significantly higher SUVR in the bilateral orbital gyri and inferior and middle frontal GM and nearby WM regions than the controls.In contrast, 4RT-like patients with bvFTD (P04-P07) showed significantly higher SUVR mainly in the bilateral superior parietal lobules, right postcentral and angular gyri, nearby WM regions, and left middle frontal cortex.
In the 3RT-like bvFTD cases, regions encompassing the right orbital gyrus remained significant after applying a cluster-level threshold of pFWE < 0.05.Meanwhile, no regions remained significant with pFWE < 0.05 in the 4RT-like bvFTD patients, presumably due to the focal distribution of tau lesions with little spatial overlap among the patients.