Tau epicenter identification and connectivity in clinically heterogeneous Alzheimer’s disease variants

Background While the spread of tau throughout the brain is well‐characterized in amnestic Alzheimer’s disease (AD), tau staging remains less clear for those suffering from clinically heterogeneous variants including logopenic variant primary progressive aphasia (lvPPA) and posterior cortical atrophy (PCA). We created a method to identify local maxima of tau SUVR (“epicenters”) in participants with early‐stage clinically heterogeneous AD with the goal of classifying participant subtypes based on the functional connectivity of these epicenters. Method We obtained 18F‐MK‐6240 tau PET scans from 10 participants with diagnoses of early‐stage clinically heterogeneous AD (5 lvPPA, 5 PCA, Clinical Dementia Rating 0.5‐1.0). We identified three local maxima of tau SUVR with a minimum specified distance of 20mm between them for each participant (Fig. 1). We then derived each epicenter’s functional connectivity in 1000 healthy controls using Neurosynth and sorted participants into subtypes based on the overlap of these functional connectivity maps. Result Participants with lvPPA showed highest tau burden in the posterior temporoparietal and temporoocciptal regions, left side predominant, and participants with PCA displayed highest tau burden in the bilateral parietooccipital cortex, frontal eye fields, and fusiform gyrus. We identified nine distinct subtypes of functional connectivity and most participants’ epicenters participated in more than one subtype. All subtypes highlight functional connectivity networks that have been associated with language and, or visuospatial processing. The most common subtype (seen in 7 of the 10 participants and 11 of the 30 epicenters) features connectivity within the bilateral occipital lobule, fusiform gyrus, and superior parietal lobule (Fig. 2). Conclusion We used a novel method to identify tau “epicenters” in the brains of participants with early‐stage clinically heterogeneous AD. Using these epicenters, we derived functional connectivity maps, which we then grouped into nine subtypes across participants. We will shortly begin collecting longitudinal tau PET in these clinically heterogeneous individuals with AD, allowing us to assess the extent to which tau spreads along the functional connectivity maps predicted by each participant’s epicenters and whether individually tailored epicenters outperform subtype classification. This work could serve as the basis for patient‐specific biomarkers of expected disease spread in clinically