[18F]-AV-1451 is a PET tracer used to evaluation of neurofibrillary tau pathology in vivo. It is still unknown what spatial pattern best characterizes an Alzheimer’s signature of tau PET uptake in the brain. The objective of this study was to utilize a data-driven method to derive a summary measure of PET tau deposition as measured by [18F]-AV-1451 for staging preclinical AD and to derive a cut-off for tau positivity.


      Participants were drawn from ongoing studies at Washington University in St. Louis. 51 cognitively normal (CN) subjects (with Clinical Dementia Rating (CDR) =0) and 12 cognitively impaired (CDR > 0, 8 CDR 0.5, 3 CDR=1, and 1 CDR=2) participants underwent PET imaging with AV-1451. Standardized uptake value ratios (SUVRs) normalized to the whole cerebellum and partial volumes corrected were calculated for 36 regions of interest (ROIs) generated using FreeSurfer. A bootstrapped sparse k-means analysis was done on ROI SUVRs from CN participants to cluster them into two groups (k=2). Unlike a normal k-means, sparse k-means cluster utilizes regional weights to determine the ROIs that maximally influenced this clustering. 45 CN and 10 CDR > 0 underwent a comprehensive neuropsychological battery and 50 CN and 12 CDR >0 underwent beta-amyloid PET imaging and were categorized as beta-amyloid positive or negative.


      The highest-weighted ROIs (Figure 1) were: entorhinal cortex, amygdala, lateral occipital cortex, and inferior temporal cortex. The unweighted mean SUVR across these four ROIs correlates with episodic memory, attentional control, and visuospatial memory in both CNs and CDR>0 participants (Figure 2). Amyloid PET positive individuals have significantly higher mean SUVRs across these ROIs (Figure 3). The midpoint between tau negative and positive group means was 1.23. Using this cut-off, tau positive CN participants had significantly higher amyloid PET burden (Figure 4).


      The unweighted mean SUVR of the entorhinal cortex, amygdala, lateral occipital cortex, and inferior temporal cortex ROIs provides a summary measure for PET Tau deposition that is sensitive to early amyloid PET deposition and early cognitive change. In these regions a cut-off SUVR of 1.23 marks the transition between PET tau positive and negative in CNs.
      Figure thumbnail fx1
      Figure 1Mean weight for ROIs across 500 simulations of sparse k-means algorithm when clustering CN participants into k=2 groups
      Figure thumbnail fx2
      Figure 2aRelationship between mean AV-1451 SUVR and episodic memory
      Figure thumbnail fx3
      Figure 2bRelationship between mean AV-1451 SUVR and visual spatial memory
      Figure thumbnail fx4
      Figure 2cRelationship between mean AV-1451 SUVR and attentional control
      Figure thumbnail fx5
      Figure 3Mean AV1451 SUVR in PET Amyloid Positive vs. Negative Participants
      Figure thumbnail fx6
      Figure 4Difference in AV45 mcSUVR in Tau Positive vs. Tau Negative CN Participants