Neuropathological correlates of age and sex differences in 18F-flortaucipir PET

Among amyloid-positive individuals with symptomatic Alzheimer's disease, older age and male sex have been associated with a lower prevalence of Tau-PET-positivity. Whether Tau-PET-negative older and/or male individuals truly do not harbor widespread tau pathology or whether tangle density is below the PET-detection threshold remains unknown. Therefore, we aimed to investigate the neuropathological correlates of age- and sex-differences in Tau-PET in independent PET-only, autopsy-only, and PET-to-autopsy cohorts. In the PET-only cohort, we included amyloid-β-positive participants with MCI or dementia who underwent [18F]flortaucipir-PET (n=672). In the autopsy-only cohort, we included participants with moderate-to-frequent CERAD scores and MCI or dementia with available data on Braak stage and tangle density (n=945). In the PET-to-autopsy cohort, we included participants who underwent antemortem Tau-PET and had undergone a postmortem assessment of Braak staging (n=85) (median PET-to-post-mortem-interval: 2.6 months). A subset additionally had tangle density data available (n=63). Tau-PET SUVr was calculated in a temporal meta-region, and Tau-PET-positivity was defined using a predefined threshold of 1.36 SUVr. Autopsy cases were categorized as Braak 0-IV versus Braak V-VI. In PET-only-analyses (age: 71.9±8.2, 52.8% male), older age and male sex were associated with a lower prevalence of Tau-PET-positivity and lower Tau-PET SUVr (all p<0.05). In autopsy-only-analyses (age: 82.7±7.9, 54.6% male), older age and male sex were associated with a lower prevalence of Braak-V/VI neuropathology (both p<0.05). Among Braak-V/VI autopsy cases (n=598), older age was associated with lower tangle density (β=-0.38, p<0.001). In PET-to-autopsy-analyses (age: 81.7±9.2, 52.9% male), Tau-PET showed excellent specificity for detecting Braak-V/VI neuropathology (100% across age-stratified and sex-stratified models), while the sensitivity decreased at older age (<83y: 92% [95% confidence interval: 80%-100%] vs ≥83y: 42% [23%-62%]) and in males (females: 85% [69%-96%] vs males: 48% [28%-68%]). Older and male participants with Braak-V/VI neuropathology showed both lower Tau-PET SUVr (age: β=-0.46, p=0.003; sex: β=-0.97, p=0.001), and in the same individuals, older participants showed trend-level lower tangle density (β=-0.31, p=0.053). The lack of age/sex-interactions indicate that the relationship between Tau-PET and tangle density is consistent across ages and sexes. Comprehensive and independent PET, autopsy, and PET-to-autopsy analyses demonstrate that the associations of older age and male sex with lower [18F]flortaucipir-PET uptake and positivity rates can be explained by lower tangle densities. [18F]flortaucipir-PET SUVr thus closely reflects tangle density, which accounts for the lower sensitivity of [18F]flortaucipir-PET for detecting low-density Braak-V/VI tau pathology in older individuals and males.