Identifying the topology of ¹⁸F-AV-1451 (also known as T807) PET tau images for diagnosis and prognosis in neurodegenerative disorders

Objectives ¹⁸F-AV-1451 is a PET tracer in development for assessment of neurofibrillary tau pathology. Understanding the relationship of tau pathology to disease stage or exploring longitudinal changes in AV-1451 uptake may require quantitation of both the amplitude and extent of tracer distribution.

Methods ¹⁸F-AV-1451 data were acquired 80-100 min post injection in 154 subjects [14 young healthy controls (YCN), 34 healthy older controls (OCN; 3 Aβ+), 72 MCI (37 Aβ+) and 34 AD (22 Aβ+)]. SUVR was calculated on a voxel-wise basis and for AAL VOIs across neocortical and mesial temporal lobe regions. Quantitative measures from PET images were evaluated for diagnostic group separation and for correlation to cognitive scores (ADAS cog and MMSE). Evaluations were conducted for VOIs and voxel-wise SUVR-thresholded images (relative to YCN). Voxel-wise correlation images were obtained for SUVR versus ADAS-cog or MMSE. Voxel-wise methods employed both an intensity-based and an extent-based metric. Finally, SUVR images were entered into a multi-block barycentric discriminant analysis (MUBADA) to identify regions showing maximal diagnostic group separation (OCN, MCI AD and Aβ+ vs. Aβ- subgroups).

Results Analyses showed mean differences between diagnostic groups for all approaches, though extent provided greater group differences (effect size in temporal lobe for distinguishing cohorts was ~2.5 for extent metrics and ~1.8 for intensity). MMSE was more strongly correlated to posterior regions than other brain areas, while ADAS-cog was more strongly correlated to both temporal and frontal regions. MUBADA identified 2 dimensions that explained 98% of the variance in regional SUVr pattern between groups, and the first of these characterized Aβ+ MCI and AD relative to all other groups.

Conclusions Most metrics were effective at demonstrating group differences. Discriminant analyses suggested temporal lobe involvement was related to aging, while frontal and parietal lobe involvement was related to multi-biomarker pathology.