COVARYING PATTERNS OF FUNCTIONAL CONNECTIVITY WITH AMYLOID AND TAU DEPOSITION IN EARLY STAGE ALZHEIMER’S DISEASE

The spatial relationships between functional connectivity (FC) and AD imaging markers of protein deposition are still not well understood. We use multi-modal connectivity independent component analysis (hybrid-connICA) to assess these relationships. We assessed resting-state-fMRI, F18-Flortaucipir (tau), and F18-Florbetapir (amyloid) in 33 ADNI3 subjects who had Florbetapir and Flortaucipir scans (Table1). Canonical regional measures in AD PET, cortical amyloid SUVR and medial temporal lobe (MTL) tau SUVR were quantified (Freesurfer6.00 regions). FC matrices were estimated (286 region parcellation). This parcellation was also applied to PET SUVR images. FC and PET vectors were merged (FC-Florbetapir, FC-Flortaucipir). connICA (5ICs) was performed on combined subject data (Fig.1), resulting in traits containing a FC-pattern and a PET-pattern along with corresponding weights, quantifying individual-subject expression of the traits. Traits were required to exhibit significant PET-FC pattern spatial correlation (p<0.05), to focus only on co-varying patterns. Traits were identified as relevant if corresponding weights correlated (p<0.01) with corresponding canonical regional SUVR measures or if weights exhibited a diagnostic effect (ANOVA, p<0.01). Relevant nodes and connections (>97.5% PET-pattern OR >97.5% FC-pattern) were visualized (Fig.2&4). One FC-Florbetapir (Fig2.A-D, Fig.3) and one FC-Flortaucipir IC (Fig.4A-D, Fig.5) met criteria. The FC-patterns correlated 0.76 and PET-patterns correlated 0.48. Both FC-patterns resembled canonical resting-state networks (Fig.2-3F). In the FC-Florbetapir pattern (Fig.2), relevant nodes consisted primarily of within/between DMN connections. In the FC-Flortaucipir pattern (Fig.4), relevant nodes primarily consisted of within/between visual-network connections.