THE RELATIONSHIP BETWEEN TAU PET AND OTHER AD BIOMARKERS IN AUTOSOMAL DOMINANT ALZHEIMER DISEASE

2017). The aim of this study was to develop a simple, quantitative tau PET-based patient classification algorithm based on sequential lobar accumulation of flortaucipir signal. Methods: The EXPEDITION3 trial enrolled amyloid-positive patients with mild AD (MMSE 20-26). Flortaucipir images were acquired for a subset of participants at baseline, week 40 and week 80. We quantified tau burden in baseline scans from the placebo arm (N1⁄497), using average SUVR values in atlas-based lateral temporal, parietal and frontal lobes with respect to a white matter-based reference region (PERSI). Elevated tau burden in the temporal (T+), parietal (P+) and frontal (F+) lobes bilaterally was determined using a single positivity threshold. All possible lobar profiles were grouped into four stages 0-III (T-P-F-, T+PF-, T+/-P+Fand T+/-P+/-F+), reflecting the stereotypical patterns of tau spread inferred from neuropathological studies. A mixed effects repeated measures method was used to characterize changes in global weighted SUVR (MUBADA), cognition and structural MRI. Results:Lobar stages were strongly concordant with a global tau burden measured by MUBADA SUVR. On average, individuals belonging to more advanced lobar stages at baseline demonstrated numerically more rapid cognitive (MMSE, ADAS-Cog14 and iADRS) and neurodegenerative (whole brain and whole temporal lobe volumes) decline over 80 weeks follow-up. In these analyses, approximately 25-30% of the cases were classified to each of stages I-III, with a slightly lower proportion assigned to stage 0. If scans were staged using a more rigorous sequential classification (T-P-F-, T+P-F-, T+P+Fand T+P+F+), <5% scans remained unclassified. Conclusions: We propose a simple quantitative algorithm based on suprathreshold lobar flortaucipir SUVR values for AD staging in vivo. Specifically, a symptomatic, amyloid-positive AD population may potentially be further segmented into four flortaucipir-measured pathological stages characterized by an escalating risk of cognitive decline. Further validation of these findings using independent datasets is warranted.