Longitudinal hippocampal atrophy is associated with an amyloid‐independent entorhinal tauopathy and an amyloid‐dependent neocortical tauopathy

Abstract

AbstractBackgroundHippocampal atrophy is observed in Alzheimer’s disease (AD), but also in older adults with no evidence of amyloid‐β (Aβ) plaques. To understand the pathophysiology of hippocampal volume (HV) loss, we investigated the associations between longitudinal HV, age, genotype, Aβ, and tau in clinically normal (CN) participants from the Harvard Aging Brain Study.MethodSerial MRI (HV [1.3‐7.0y], PiB‐PET (Aβ, [1.9‐8.5y]), and Flortaucipir‐PET (tau, [0.8‐6.0y]) measures were obtained from 128CN participants (72 [56%] females, 38 [30%] e4 carriers, median age at baseline: 71.3, follow‐up duration: 5.1y). Participants had a median of 3MRIs [2‐5], 3PiB‐PET [2‐5], and 2Flortaucipir‐PET [2‐4]. Longitudinal HVs were processed using Freesurfer v.6 and adjusted for intracranial volume. PiB was measured in a neocortical aggregate, Flortaucipir in inferior temporal (IT) and entorhinal cortex (EC). PET data were expressed as PVC‐SUVr scaled to subcortical white matter. We predicted imaging data over time with random intercept and slope in linear mixed‐models and extracted PiB, FTP, and HV slopes for each subject. Baseline PET and slope data were entered in age‐adjusted linear regressions to evaluate their associations with HV slope.ResultFaster HV loss was observed at older ages (Figure 1), and marginally in e4 carriers (Table, #1‐2). It was also associated with higher baseline PiB levels (#3). The PiB association was stronger than the one of e4 status (#3). HV slope did not correlate with PiB slope (#4). HV loss was associated with baseline EC‐FTP (#5) and baseline IT‐FTP (#6). The association between HV loss and EC‐FTP was also observed in the low‐PiB only (#7), but not the one with IT‐FTP (#8). IT‐FTP, but not EC‐FTP, interacted with baseline PiB to predict HV slope (Figure 2‐3). FTP slopes measures were associated with HV slope, above and beyond the baseline association (#9‐10). Altogether PET data explained 40% of the variance in HV slope; but faster HV loss was still associated with age after taking PET measures into account.ConclusionIn preclinical AD, hippocampal atrophy is associated with an Aβ‐independent entorhinal tau accumulation and an Aβ‐dependent neocortical tau accumulation. It is also likely associated with pathological processes that our biomarkers did not measure.