Different relationship of cross‐regional correlations of progression of local cortical thickness and local SUVR 18F‐flortaucipir PET values to strength of connectivity: Development of new models and analysis methods/tau.

Background: We have demonstrated in vivo in humans that spatial progression of tau distribution and atrophy in Alzheimer's is largely driven by connectivity (Soucy et al, HAI2020). Nevertheless, local tau accumulation and atrophy should not progress in parallel during disease evolution. Atrophy keeps worsening, while local tau density (SUVRs estimated with PET) initially increases but could then decline because of tissue loss. Therefore, despite the impact of connectivity on progression of both local atrophy and SUVRs, it likely will not remain the same for both over time. This is what we tested in this work. Method: We studied 35 AD subjects from ADNI with 2 PET studies (18F‐florbetapir) at times Ta/Tb (mean elapsed time = 438 days, median = 454), and a T1 MRI at Ta. Eighteen subjects were non‐progressors (NP) and 17 were progressors (P) based on change between Ta/b hippocampal SUVRs values. This criterion was considered to represent how far advanced patients were in their disease. Moran I statistic was used to measure autocorrelations between connectivity and T1 MRI‐defined cortical thickness at Ta or tau PET SUVR values at both times Ta and Tb in NP and P subjects in regions separately analyzed based the "intensity" of their connectivity as defined by a standard connectivity map (Human Connectome Project). We defined 3 groups of regions from thresholds separating poorly connected from moderately connected (0.01) and moderately connected from highly connected ones (0.021). Result: In both NPs and Ps, regions with high connectivity showed higher Z‐score (4.06) of the Moran I value for correlations between regions' cortical thickness than those with low (2.64)/moderate (2.17) connectivity. For SUVR values however, in both NPs and Ps and at both time points, Z‐scores were always lower for the highly connected regions than for those with moderate/low connectivity. Conclusion: The way connectivity influences correlations of atrophy across regions is indeed different from the way it affects correlations of tau deposition, likely for the reasons we hypothesized above to guide the progression of each parameter. Models of tau propagation across the brain will need to factor in ongoing atrophy but reversible accumulation of tau. [ABSTRACT FROM AUTHOR]