P2‐360: COVARYING SPATIAL PATTERNS OF TAU DEPOSITION AND GRAY MATTER ATROPHY UNEARTHED BY THE INFORMED MULTIMODAL PARTIAL LEAST SQUARES (MMPLS) IN AUTOSOMAL DOMINANT ALZHEIMER'S DISEASE: FINDINGS FROM THE COLBOS PROJECT

Abstract

We have shown that presenilin-1 (PSEN1) mutation carriers from the Colombian kindred with autosomal dominant Alzheimer's disease (ADAD) have elevated tau levels 6–10 years before their estimated clinical onset, which are associated with cognitive decline (Quiroz et al., 2018). However, the interaction between tau pathology and neurodegeneration as measured by brain atrophy is unclear in this population. In this study, we investigated this interaction in cognitively unimpaired carriers and noncarriers from this kindred using an informed multimodal partial least squares (MMPLS) approach, previously developed to jointly examine hypometabolism and brain atrophy in late-onset Alzheimer's disease (Chen et al., 2009, 2012). Flortaucipir PET and sMRI images were analyzed from 32 members of the Colombia-Boston Longitudinal Biomarker Study (COLBOS); 14 unimpaired carriers and 18 age-matched noncarriers. Tau PET images were coregistered to the T1 MRI. T1 images were warped into template space and the warp was applied to coregistered tau PET images. The cerebellum crux was used as reference region for PET. Gray matter density maps were calculated. Voxel-wise Informed MMPLS was applied to gray matter density and tau PET images of the unimpaired carriers and noncarriers as well as SPM12 voxel-wise univariate analyses for each gray matter density and PET separately. Using a univariate approach, PSEN1 mutation carriers had higher levels of tau binding throughout the lateral parietal, temporal cortices, precuneus, and medial temporal regions. There were not significant differences in levels of atrophy between carriers and noncarriers. In contrast, MMPLS showed that covarying elevated tau levels were confined to medial regions (precuneus, medial temporal), while the covarying pattern of atrophy seemed to spread to parietal, left medial temporal, and precuneus. Further, the elevated tau levels in medial regions corresponded to areas with atrophy. Compared to commonly used univariate analyses, a multivariate approach provides enriched information of the interaction of the spatial patterns of tau burden and atrophy in preclinical stages of ADAD. Longitudinal analyses in ADAD with larger samples are needed to further investigate the interaction of tau pathology and atrophy in the clinical progression of the disease, from asymptomatic to symptomatic stages.