Tau phosphorylation is more closely associated with amyloid-β plaques than with tau neurofibrillary tangles

Abstract

Background

While fluid phosphorylated tau (pTau) epitopes are interpreted to be biomarkers of tau pathology according to the A/T/(N) framework, it is unclear to what extent they are preferentially associated with the defining histopathological hallmarks of Alzheimer’s Disease (AD): amyloid-β plaques and tau neurofibrillary tangles.

Method

We studied 171 individuals, including young adults (n=27), cognitively unimpaired elderly (n=85), individuals with mild cognitive impairment (n=36) and individuals with Alzheimer’s clinical syndrome (n=23), who were evaluated with [18F]AZD4694 amyloid-PET, [18F]MK6240 tau-PET, four tau phosphorylation sites in CSF (pTau181, pTau217, pTau231, pTau235) and two phosphorylation sites in plasma (pTau181, pTau231). To better understand the role of soluble biomarkers in AD diagnostic and research purposes, we evaluated associations between soluble pTau sites and cerebral amyloid-PET and tau-PET concentrations. Voxel-wise linear regressions between fluid and imaging biomarkers were performed. Results for plasma and CSF pTau181 were replicated in an independent sample of 258 individuals included in the Alzheimer’s Disease Neuroimaging Initiative cohort (ADNI).

Results

We observed that for all plasma and CSF epitopes, pTau is more closely associated with amyloid-PET than with tau-PET. Squared correlation coefficients (Spearman’s R2) between CSF pTau epitopes and neocortical [18F]AZD4694 SUVR range from 0.48 to 0.63, while those for the temporal meta-ROI [18F]MK6240 SUVR vary between 0.33 and 0.45 (p<0.001 for all phosphorylation sites). Using the R package “Cocor”, for CSF pTau181, pTau217 and pTau231 the difference in correlation coefficients between both measures of imaging biomarkers appear significant (p<0.01). Voxel-wise linear regression analyses further support these results, in particular showing sizable associations for CSF pTau231 with amyloid-PET. In addition, soluble pTau is more closely correlated with medial temporal than with neocortical tau. All soluble pTau concentrations rise significantly with increasing amyloid-β plaque load. In contrast, soluble pTau plateau as tau-PET concentrations increase, starting at Braak stage III. These findings were replicated using plasma and CSF pTau181 in the ADNI cohort.

Conclusion

Phosphorylated tau epitopes measured in CSF and plasma better reflect cerebral amyloidosis than neurofibrillary tangles in the brain. The current findings support careful interpretation of fluid pTau concentrations when implementing the A/T/(N) framework.