Differential Associations of Microglial Inflammation on LATE-NC and Tangle-Related Hippocampal Atrophy
BACKGROUND Accumulations of AD and LATE-NC both contribute to changes in hippocampal volume, possibly via distinct and/or overlapping mechanisms. Microglia-driven inflammation is a shared pathway associated with both AD and LATE-NC. However, the extent to which microglia inflammation is associated with hippocampal volume is less understood. OBJECTIVE Examine the relationship between AD and LATE-NC with hippocampal volume in persons with differing levels of microglia inflammation. METHODS Cerebral hemispheres from 441 older adults who came to autopsy were studied. All hemispheres underwent ex-vivo MRI and detailed neuropathologic examination for neurodegenerative and cerebrovascular pathologies. Microglia were quantified in the hippocampal CA1/subiculum region using machine learning-based classifiers trained on digitized CR3-43-stained images via the HALO digital pathology platform. First, linear regression models examined the association of microglia with hippocampal volume, adjusting for demographics, postmortem interval (PMI), and common age-related pathologies. Second, linear regression models were employed to examine whether microglia density modified associations of β-amyloid, tangle, or LATE-NC on hippocampal volume. RESULTS Participants had a mean age of 90 years at death with 75% being women. Intermediate or high likelihood ADNC was present in 64% and LATE-NC (stage 2/3) was present in 52%. In linear regression models, adjusting for demographics and PMI, higher microglia density was associated with a lower hippocampal volume to hemisphere ratio (estimate = −0.021 SE=0.01, p=0.002); however, after adjusting for common age-related pathologies the association was attenuated (p=0.70). β-amyloid, tangles, and LATE-NC remained independently associated with a lower hippocampal volume. The association of LATE-NC with hippocampal volume was stronger in brains with greater microglia burden (estimate for the interaction term = −0.016; SE=0.01, p=0.002). No interactions were seen between β-amyloid or tangles with microglia on hippocampal volume. In stratified analyses, microglial density modified the association between LATE-NC and hippocampal volume, independent of AD neuropathologic status. CONCLUSION Microglia-driven inflammation strengthens the association of LATE-NC, but not AD pathology, on hippocampal volume loss. These findings emphasize the importance of inflammatory pathways [when interpreting MRI-based neurodegeneration markers] in aging and mixed pathology.