Advertisement

RELATIONSHIP BETWEEN TAU POSITRON EMISSION TOMOGRAPHY WITH [18F]-AV-1451 AND LONGITUDINAL CORTICAL ATROPHY IN ALZHEIMER DISEASE

      Background

      Neurofibrillary tau pathology is a marker of neurodegeneration and can be evaluated using the PET tracer [18F]-AV-1451 (flortaucipir, T807). Many studies have added tau imaging to ongoing longitudinal cohorts. We wanted to evaluate whether longitudinal MRI scans could predict tau PET positivity in preclinical and symptomatic Alzheimer disease (AD).

      Methods

      87 cognitively normal (with CDR=0) and 14 cognitively impaired (CDR> 0, 9 CDR 0.5, 3 CDR 1, and 2 CDR 2) participants were drawn from studies on aging at Washington University in St. Louis. Participants had one or more MRI sessions preceding a visit where they acquired both a MRI scan and underwent PET imaging with AV-1451, with mean follow-up from first MRI of 5.3 (sd 2.3) yrs. A subset (n = 93) also underwent florbetapir beta-amyloid imaging. MRIs were processed using FreeSurfer to generate mean cortical thickness in each region of interest (ROI). PET data was converted to standardized uptake value ratios (SUVRs) normalized to the cerebellum and partial volume corrected. Global tau burden was estimated by the mean SUVR from entorhinal cortex, amygdala, inferior temporal cortex, and lateral occipital cortex ROIs. For each person, a slope estimating structural atrophy in each ROI was quantified by fitting all longitudinal MRI measurements in a generalized linear model (GLM). These slope estimates were then used to predict tau burden in a GLM while controlling for baseline age and gender. Participants who also had beta-amyloid imaging were fit into a second GLM, with an additional covariate of florbetapir mean cortical SUVR. Multiple comparisons were controlled using a false discovery rate.

      Results

      Antecedent cortical thinning was significantly associated with tau deposition throughout the cortex in the entire cohort (Figure 1). The effects were most prominent in the lateral temporal lobe and inferior parietal areas. These associations remained even after controlling for florbetapir levels (Figure 2) and were evident even in cognitively normal cohorts alone (Figure 3).

      Conclusions

      Antecedent cortical thinning predicts current PET Tau in preclinical AD and symptomatic AD. This relationship holds in AD even after controlling for PET beta-amyloid burden. This may be useful for participant selection for tau PET imaging or clinical trials.
      Figure thumbnail fx1
      Figure 1Relationship between antecedent cortical atrophy and current PET Tau, controlling for baseline age and gender in cognitively normal and AD participants.
      Figure thumbnail fx2
      Figure 2Relationship between antecedent cortical atrophy and current PET Tau, controlling for baseline age, gender, and current florbetapir mean cortical SUVR in cognitively normal and AD participants.
      Figure thumbnail fx3
      Figure 3Relationship between antecedent cortical atrophy and current PET Tau, controlling for baseline age and gender in cognitively normal participants.