AbstractBackgroundCSF total tau (t‐tau) is postulated as a marker of neuronal degeneration in the ATN criteria. Here, we hypothesize that CSF t‐tau is more closely related to synaptic loss than over neurodegeneration.MethodWe evaluated 128 older individuals (75 CU,33 MCI,20 AD) in the TRIAD cohort (McGill, Canada) with CSF t‐tau, SNAP25long and Neurofilament Light (NfL), Amyloid‐beta (Aβ) positron emission tomography and magnetic resonance imaging, and clinical assessments. We used increased SNAP25long as an index of synaptic dysfunction (S+) and reduced hippocampal volume as neurodegeneration (N+). S and N positivity were determined using a cutoff anchored in young CU individuals. Linear regression models were used to test the association between CSF t‐tau, SNAP25long and NfL levels and hippocampal volume, adjusted for age, sex, and cognitive status. ANCOVA was employed to inspect CSF t‐tau levels across the groups 1) S‐N‐; 2) S+N‐; 3) S‐N+; 4) S+N+, adjusted for age, sex, cognitive status, and Aβ burden.ResultWe found that a model with the addition of SNAP25long to the covariates explained 65% of the variance (p‐value<0.0001) of CSF t‐tau. Alternatively, hippocampal volume (R2 = 0.2187; p‐value<0.0001) and NfL (R2 = 0.3671; p‐value<0.0001) alone or in combination explained less than 40% of the variance of CSF t‐tau (R2 = 0.3597; p‐value<0.0001) (Figure 1). Interestingly, we observed an increase of CSF t‐tau in the S+N‐ group but not in the S‐N+, further supporting that CSF t‐tau is abnormal in the presence of synaptic dysfunction rather than neurodegeneration. Additionally, Loess regression lines confirm a larger magnitude of change of synaptic dysfunction as a function of t‐tau, compared with neurodegeneration (Figure 3). ConclusionOur results suggest that CSF t‐tau may reflect synaptic loss rather than neuronal degeneration. This has implications for the interpretability of results obtained with CSF t‐tau in biomarker studies.