Colocalization of tau PET uptake and cerebral blood flow alterations within subjects with MCI and AD
David Scott, Luc Bracoud, Chris Conklin, Madhura Ingalhalikar, Joyce Suhy- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
The accumulation of neurofibrillary tangles contributes to a neurodegenerative cascade toward Alzheimer’s disease, of which altered cerebral metabolism is also a feature. 18F‐Flortaucipir PET allows visualization of tau pathology in vivo, while MRI‐based arterial spin labeling techniques assess metabolism via cerebral blood flow. We sought to characterize the within‐subject relationship between tau burden and cerebral blood flow throughout the brain.
Method
130 ADNI3 subject visits with concurrent MRI ASL and flortaucipir PET scans were selected from the database (http://adni.loni.ucla.edu) for analysis. ADNI PET core PVC values referencing inferior cerebellar cortex were used to index SUVR across the entire FreeSurfer parcellation, and ADNI MRI core regional CBF values were identified from an overlapping set of 71 FreeSurfer regions. Linear partial correlation coefficients were calculated for tau PET SUVR and CBF in each brain region, controlling for age and amyloid PET SUVR. Lasso logistic regression returned penalized maximum‐likelihood fitted coefficients for a generalized linear model estimating meta‐temporal tau SUVR from regional CBF; regions with non‐zero coefficients in the sparsest model are reported.
Result
Following correction for age and amyloid load, significant within‐subject partial correlations between tau PET uptake and ASL CBF were observed throughout the brain. The associations were strongest in amyloid positive subjects with MCI or AD. Correlation strength was regionally specific and strongly positive (marginal p < 0.005) in subcortical and early Braak regions, while strongly negative (marginal p < 0.005) across frontal and later Braak regions. The strongest tau burden associations were observed with hypermetabolism in hippocampus (p = 0.0005, partial r = 0.59), and hypometabolism in posterior cingulate gyrus (p = 0.0001, partial r = ‐0.63). Lasso regression revealed a subset of non‐zero coefficients in the sparsest model including subcortical, temporal, parietal and frontal brain regions.
Conclusion
Tau PET uptake is associated with a pattern of hypermetabolism in early Braak stages, and hypometabolism in later Braak stages. This suggests a dynamic relationship where tau accumulation and cerebral blood flow alterations colocalize, but the direction of the effect is regionally specific.