Abstract Details

Progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD) show 4-repeat tau deposition in neurons (neurofibrillary tangles and neuropil threads), oligodendrocytes (coiled bodies) and astrocytes (tufted astrocytes [PSP] or astrocytic plaques [CBD]). Determining how MRI findings relate to underlying pathology is critical to evaluate their utility as surrogate markers to aid in diagnosis and as outcome measures for clinical trials.

To determine the relationship between tau lesion burden and MRI volume loss and DTI white matter tract abnormalities in 4-repeat tauopathies.

Forty-seven patients with antemortem 3T MRI including volumetric and DTI scans and postmortem pathological diagnosis of PSP (n=28) or CBD (n=19) were included. Tau lesion types were semiquantitatively graded in disease-specific regions. Regional volumes, fractional anisotropy (FA), and mean diffusivity (MD) were correlated using linear regression models with tau lesion scores considered separately, based on cellular type (neuronal vs glial), or summed (total tau).

Greater total tau burden correlated with volume loss in the globus pallidus (p=0.004), subthalamic nucleus (p<0.001), substantia nigra (p=0.024), red nucleus (p=0.003) and midbrain (p<0.001), with glial lesions (mostly coiled bodies) driving the associations. Neuronal lesions trended to associate with volume of the precentral cortex (p=0.061). Decreased FA and increased MD in the superior cerebellar peduncle (SCP) correlated with glial tau in cerebellar dentate (p=0.037 and p=0.019, respectively) and red nucleus (p<0.001 for both). In the midbrain, glial pathology correlated with increased MD (p=0.008). Finally, increased subcortical white matter MD was related to total tau in superior frontal (p=0.020) and precentral (p=0.017) cortices.

MRI volume of key subcortical and brainstem regions, SCP degeneration and subcortical white matter abnormalities correlated with tau deposition at autopsy making them good surrogate markers of pathology. Moreover, volume correlated with glial lesions in subcortical and brainstem areas but trended to associate with neuronal lesions in cortical regions, possibly suggesting different pathomechanisms.