Neuropathological evidence that the association between APOE‐e4 genotype and higher medial temporal lobe tau burden is mediated by higher local amyloid‐β levels

Abstract Background Evidence from previous PET imaging studies suggests that increased neocortical tau seen in APOE‐ε4 carriers can be explained by the higher levels of neocortical amyloid‐β (Aβ) pathology also seen in this group. However, these studies also suggest an amyloid‐independent association between APOE‐ε4 and medial temporal lobe (MTL) tau. Using neuropathological data, we investigated whether amyloid‐β in the MTL, which is difficult to accurately measure with PET, may fully mediate the association between APOE‐ε4 carriership and MTL tau burden. Method Amyloid‐β and tau pathology in MTL and neocortical areas have been measured in a total of 2,704 individuals from two independent neuropathological datasets (Arizona Study of Aging and Neurodegenerative Disorders, Brain and Body Donation Program, n=1,117 and ROSMAP, n=1,587; Table 1). Regression models were used to investigate the associations between APOE‐ε4 carriership and neocortical and local (MTL) amyloid‐β on MTL tau, adjusting for age, sex and presence of co‐pathologies. Based on these results, we also tested whether local (MTL) amyloid‐β mediated the association between APOE‐ε4 carriership and MTL tau. Result Supporting results from previous studies, APOE‐e4 carriers showed higher tau burden in the MTL (p<0.001) when amyloid‐β was not included in the model in both datasets. In independent models, both local and neocortical amyloid‐β were associated with MTL tau, with local amyloid‐β showing a stronger association with MTL tau in both datasets (Arizona: β MTL =0.93 vs. β neoc =0.76; ROSMAP: β MTL =0.42 vs. β neoc =0.40). When both measures were included in the model, local amyloid‐β also showed a stronger association with MTL tau in ROSMAP (β MTL =0.28 vs. β neoc =0.19). In the Arizona dataset, neocortical amyloid‐β effect became non‐significant (p=0.252) when local amyloid‐β was also included in the model. Finally, with mediation analyses we observed that MTL amyloid‐β fully mediated the APOE‐ε4 effect on MTL tau in both datasets (Figure 1). All statistics are reported in Table 2. Conclusion In two large neuropathological datasets, we showed that the association between APOE‐ε4 genotype and higher medial temporal lobe tau burden is mediated by higher local amyloid‐β levels. This finding contradicts some previous imaging literature, where only global amyloid‐β burden was considered.