Combining high‐resolution ex vivo MRI and histopathology to identify medial temporal lobe atrophy patterns specific to tau pathology in Alzheimer’s Disease

Background Tau pathology in the medial temporal lobe (MTL) is closely linked to neurodegeneration, and is an early pathological change associated with Alzheimer’s disease (AD). MRI measures of MTL neurodegeneration have proven to be sensitive to change during preclinical AD. Current measures are confounded by the presence of non‐AD pathologies (e.g., TDP‐43, ageing). Here, we combine ex vivo imaging with histopathological ratings of tau and TDP‐43 to identify fine‐grained MTL atrophy patterns specific to tau. Such an analysis could be used to define MTL “hotspots” where in vivo measures of neurodegeneration are expected to be strongly associated with tau, potentially enabling the development of biomarkers that are more effective during early AD clinical trials. Method Ex vivo MRI scans (0.2x0.2x0.2mm 3 , 9.4T) of human MTL specimens were combined using a customized registration approach to construct a 3D atlas. Using serial histology available in a subset of specimens (n = 11), MTL subregions in the atlas were labelled based on cytoarchitecture (Ravikumar et al., 2021) (Figure 1A). To perform thickness analysis, 29 specimens containing a primary diagnosis of AD or primary age‐related tauopathy were registered to the atlas (13 specimens were registered to the atlas during atlas construction, and 16 were registered to the atlas after it was constructed) (Figure 1B, Table 1). Using histopathology measures of tau and TDP‐43 pathology (based on contralateral sampling), we investigated the association between tau and thickness by fitting a linear model (with age/TDP‐43 as covariates) at each point along the MTL and SRLM surface. Result Pointwise thickness analysis reveals significant atrophy patterns in the transentorhinal region and SRLM. When excluding age from the model, stronger tau associations are observed in the SRLM, entorhinal cortex, and extending further towards Brodmann Area 35 (Figure 2). Conclusion Our findings are consistent with early Braak stages but suggest that covarying for age may be obscuring some associations due to tau since age and tau burden have a highly correlated relationship (Figure 3). In future work, quantitative maps of NFT burden will be mapped from serial pathology images into atlas space, allowing us to characterize NFT distribution in 3D (Yushkevich et al., 2021).