Cortical tau deposition promotes atrophy in connected white matter regions in Alzheimer's disease.

In Alzheimer's disease (AD), fibrillar tau pathology is a key driver of neurodegeneration and cortical atrophy. Yet, emerging evidence suggests that tau aggregates also contribute to white matter (WM) damage. Specifically, physiological tau stabilizes intra-axonal microtubules, while hyperphosphorylated tau disrupts microtubule integrity, ensuing intraneuronal tau aggregation, neuronal disconnection, and axonal degeneration. Therefore, we investigated whether cortical tau promotes atrophy in connected WM regions in AD. To this end, we included 186 amyloid-positive (Aβ+) patients across the AD spectrum and 102 cognitively normal (CN) amyloid-negative (Aβ-) participants from ADNI, with baseline amyloid-PET, tau-PET, and T1-weighted MRI. Longitudinal tau-PET and MRI (∼2-years) were available for a subset of 138 participants, to further assess the relationship between tau accumulation and WM atrophy over time. For replication, we included 378/60 CN Aβ+/Aβ- participants from the A4/LEARN cohort with baseline amyloid-PET, tau-PET, and T1-weighted MRI, where a subset of 141/4 CN Aβ+/Aβ- subjects had ∼5-year longitudinal tau-PET and MRI. Cortical tau-PET Standardized Uptake Value Ratios were extracted from 210 cortical regions of the Brainnetome Atlas. In addition, we used a diffusion MRI-based tractography template to determine WM volumes of fiber tracts connected to cortical regions using segmented T1-weighted MRI. Using linear regression, we tested whether higher cortical tau-PET at baseline was associated with (i) lower baseline WM volume and (ii) faster WM volume loss over time, and (iii) whether faster longitudinal tau-PET increases paralleled faster WM loss. Testing the reverse model examined whether baseline WM atrophy predicted faster subsequent tau-PET increase in connected regions. Models were adjusted for age, sex, intracranial volume, WM hyperintensity volume, ApoE4 status and global amyloid-PET. In ADNI, elevated baseline cortical tau-PET in temporal regions was associated with lower baseline WM volume in adjacent regions, with more pronounced effects in patients across the AD spectrum and with weaker associations in the preclinical A4/LEARN sample. In both samples, higher baseline temporo-parietal tau-PET as well as faster tau-PET increase over time were significantly linked to accelerated volume loss in connected WM regions, which was especially pronounced in individuals on the AD spectrum. Importantly, baseline WM volume did not predict subsequent tau-PET change rates in adjacent cortical regions, suggesting a unidirectional relationship between fibrillar tau and subsequent WM degeneration. Together, our findings suggest that cortical tau accumulation promotes atrophy in adjacent WM regions in AD, highlighting that tau-induced axonal degeneration and potentially neuronal disconnection may play a pivotal role in disease progression.