Novel Multimodal Imaging Framework for Synaptic Integrity Characterization in Alzheimer’s Disease

Background Dynamic loss of synapses is well understood to be a crucial endpoint in many neurodegenerative diseases especially that of Alzheimer’s Disease (AD). Hence, successful in vivo characterization of synaptic integrity (SI) provides huge potential for opportunities in therapeutics targeting synaptic damage restoration. Though recent studies using synaptic glycoprotein 2A (SV2A) PET imaging remains as the most direct measure of SI, such methods are not readily available for clinical applications. Here, we examine a new framework for in vivo characterization of SI in the AD continuum by incorporating widely available imaging parameters such as atrophy, connectivity, and neuronal metabolism. Method 53 amyloid‐negative cognitively normal (CN‐), 92 amyloid‐positive mild cognitive impairment (MCI+), and 50 amyloid‐positive AD (AD+) subjects with T1‐weighted MRI, diffusion‐weighted MRI and 18F‐fluorodeoxyglucose (FDG) PET images were collected from ADNI. Selected imaging modalities were each theorized to portray distinct parts of SI such as pre‐synaptic disruption, synaptic connectivity, and synaptic dysfunction. Hence, using an exogenous common factor analysis model, with each modality measures as primary manifestation variables, an amalgamated regional SI score was calculated. Using regional values, factor analysis was implemented again to produce global SI scores for each subject. Regional scores were normalized and adjusted for age, sex, and education for groupwise comparison. Result Regional SI W‐scores showed progressive decline along the AD continuum in the entorhinal, lateral temporal, lateral parietal, and isthmus cingulate cortices, showing significant differences among most regions between CN‐ and AD+ groups. Bilateral entorhinal and isthmus cingulate cortices and left inferior temporal region showed strong significant differences between CN‐ and MCI+ groups, suggesting early synaptic loss in these areas, which agree well with previous results from SV2A PET studies. Global scores also declined significantly through disease progression and showed associations with decreased cognitive performance as measured by MMSE and CDR‐SOB. Conclusion Various neuroimaging biomarkers have been extensively verified to observe different traits of AD, yet successful characterization of SI without SV2A PET is first to our knowledge. Our model has shown that synaptic change can be indirectly visualized using widely available imaging modalities, providing immense opp