Tau pathology mediates age effects on medial temporal lobe atrophy

Background Age effects on the hippocampus are well established, but age‐effects on adjacent cortical regions less so. Moreover, aging has often been investigated using cross‐sectional designs, seldomly separating aging effects from the effects of β‐amyloid. We investigate the effect of aging on medial temporal lobe (MTL) subregions cross‐sectionally and longitudinally in β‐amyloid‐negative (A‐) cognitively normal (CN) older adults and the role of tau pathology, a potential mediator. Method T1‐MRI scans of 191 A‐ CN adults (71.7±6.0 years, 52.4% male) from ADNI‐GO/2 were segmented using a specialized method for MTL morphometry to obtain anterior/posterior hippocampus volumes, entorhinal cortex (ERC), Brodmann areas (BA) 35/36 and parahippocampal cortex thickness. Annualized atrophy rates were estimated with an unbiased deformation‐based morphometry method using follow‐up MRI scans within 2.1±0.1 years. All MTL subregions were corrected for sex and hippocampal volume also for intracranial volume. Cerebrospinal fluid levels of phosphorylated tau (p‐tau) were logtransformed. β‐amyloid status (A+/‐) was determined using a standard cutoff for amyloid PET. Result Cross‐sectionally age was significantly associated with hippocampal volumes and BA35 (Figure 1‐2). Longitudinally, MTL subregions showed atrophy rates significantly smaller than 0, and paired t‐tests revealed a significantly faster atrophy rate for BA35 than the other cortical regions (Figure 1). Older age was associated with increased atrophy rates in hippocampal volumes, ERC and BA35 (Table 1, Figure 3). A significant association was found for p‐tau levels with anterior and posterior hippocampus (trend) and with anterior and posterior hippocampal atrophy rates (Figure 4), which remained similar after a correction for age. Since the association for anterior/posterior hippocampus was similar, we performed a mediation analysis for total hippocampus, showing that aging leads to smaller hippocampal volumes (p<0.05) and faster hippocampal atrophy rates (p<0.10) through higher p‐tau levels (Figure 5), while covarying for β‐amyloid levels. Conclusion The cross‐sectional age effects on the MTL roughly matched longitudinal atrophy rates in A‐CN older adults, with most pronounced atrophy in the hippocampus and BA35, early sites of neurofibrillary pathology. The mediation analyses further supported an important role for tau pathology suggesting that Primary Age‐Related Tauopathy may partly mediate the observed age effects on the MTL.