Topographic distribution of MRI‐visible enlarged perivascular space (ePVS) in Alzheimer's disease: Role of ePVS located at the temporal lobe

Background Accumulating data suggests that MRI‐visible enlarged PVS (ePVS) may serve as an imaging marker of glymphatic dysfunction. Until now, ePVS in various regions such as basal ganglia (BG‐ePVS) or centrum semiovale (CS‐ePVS) have been introduced and their topographic distributions are recognized as a marker to differentiate the underlying etiologies, although they failed to predict amyloid‐positivity. As the conventional classification of ePVS was insufficient to predict Alzheimer pathologies, we developed a new grading system for ePVS located at the temporal lobe (T‐ePVS), a strategic region where both amyloid and tau overlaps in their early stages. We investigated whether amyloid‐positivity or tau‐positivity is associated with T‐ePVS along with other conventional ePVS markers. Method ePVS were manually counted on magnetic resonance (MR) images obtained from 272 participants (cognitively normal, n=96; cognitively impaired, n=176) who underwent dual PET scans (18F‐flortaucipir and 18F‐florbetaben). T‐ePVS and BG‐ePVS were defined as “extensive” when the counts in any hemisphere were >10, while the cutoff of CS‐ePVS was >20. Subjects were grouped based on the combinations of the severity of region‐specific ePVS and their relationship with amyloid‐/tau‐positivity were investigated. Moreover, logistic regression models were constructed to investigate whether severity of region‐specific ePVS were predictive of amyloid‐ or tau‐positivity. Result Compared to those with low T‐ePVS burden (21/136, 15.4%), subjects with high T‐ePVS burden were more likely to show amyloid‐positivity (89/136, 65.4%). Among the 8 groups based on region‐specific ePVS burden, subjects with high T‐/CS‐ePVS but low BG‐ePVS burden showed highest amyloid‐positivity (46/56, 82.3%). Furthermore, high T‐ePVS burden (Odds ratio [OR]=7.458; 95% confidence interval [CI]=3.387‐16.419), low BG‐ePVS burden (OR=0.238; 95% CI=0.108‐0.524), and APOE ε4 allele presence (OR=5.175; 95% CI=2.444‐10.959) were significantly associated with amyloid‐positivity. Similar results were reproduced when the analyses were performed within subjects with cognitive impairment. Within the amyloid‐positive subjects, subjects with high CS‐ePVS but low BG‐ePVS burden were likely to exhibit tau‐positivity, while the severity of each region‐specific ePVS were not independently predictive of tau‐positivity. Conclusion Our findings suggest that investigating the load and topographic distribution of ePVS based on MR images