Increased CSF GAP‐43 is associated with accelerated tau accumulation and spread in Alzheimer’s disease

Abstract

AbstractBackgroundAnimal and in vitro studies of Alzheimer’s disease (AD) found that tau spreads trans‐synaptically in an activity‐dependent manner, suggesting that synapses are key routes for tau spread. Importantly, amyloid‐beta (Ab) induces synaptic remodeling and aberrant synaptic activity, which may accelerate trans‐synaptic tau spread. In AD patients, we found previously that PET‐assessed tau spreads from circumscribed epicenters across functionally connected brain regions. Yet, it remains unclear whether connectivity‐associated spread of tau in AD is accelerated by Ab‐related synaptic changes. The presynaptic growth‐associated protein 43 (GAP‐43) is implicated in synaptic plasticity and is consistently increased in cerebrospinal fluid (CSF) in AD, suggesting that GAP‐43 captures deviations in synaptic integrity such as aberrant synaptic activity. We therefore tested whether higher GAP‐43 is associated with faster Ab‐related tau spread.MethodsWe included longitudinal [18F]Flortaucipir tau‐PET (follow‐up∼3.21±1.48yrs), baseline [18F]Florbetapir amyloid‐PET and CSF GAP‐43 from 93 subjects (i.e., cognitively normal [CN] controls Ab‐, n = 54; CN/Mild Cognitive Impairment Ab+, n = 33/21). To model connectivity‐associated tau spread, we determined a connectivity template across 200 cortical regions (i.e., Schaefer atlas) using 3T multi‐band resting‐state fMRI in 42 age‐ and sex‐matched Ab‐/tau‐ controls. All statistical models were controlled for age, sex and diagnosis.ResultsUsing linear regression, we found that higher CSF GAP‐43 levels were associated with faster Ab‐related tau‐PET increase in pre‐defined ROIs (centiloid x GAP‐43 interaction, global/temporal‐meta‐ROI: b = 0.0019/0.0028; p = 0.009/0.007, Fig.1) as well as in the personalized Q1 ROI (b = 0.0024; p = 0.004) that summarizes regions most closely connected to subject‐specific tau epicenters with highest baseline tau‐PET (Fig.2A). Importantly, this centiloid x GAP‐43 interaction decreased across regions less strongly connected to tau epicenters (Q2/Q3/Q4: b = 0.0017/0.0008/0.0002; p = 0.009/0.080/0.344, Fig.2B‐D, 3A). Further, we quantified subject‐specific connectivity‐associated tau spread (i.e., the regression‐derived association between epicenter connectivity and tau‐PET increase in remaining brain regions). Here, higher CSF GAP‐43 was associated with a stronger association between Ab and connectivity‐associated tau spread (centiloid x GAP‐43 interaction, b = ‐0.0031; p = 0.037, Fig.3B).ConclusionsHigher CSF GAP‐43 is associated with faster Ab‐related tau spread across interconnected brain regions in AD, suggesting that AD‐related synaptic abnormalities are associated with accelerated tau spread rendering synaptic changes a potential treatment target to attenuate tau spreading.