Comparative Evaluation of SNAP25 in Cerebrospinal Fluid and Plasma Across Three Cohorts

Abstract Background Synaptic dysfunction is a pathophysiological feature of Alzheimer’s disease (AD). Although levels of synaptosomal-associated protein 25 (SNAP25) in cerebrospinal fluid (CSF) have been shown to be elevated, the utility of plasma SNAP25 and its relationship with synaptic density remains unclear. We evaluated and compared plasma and CSF SNAP25 across three independent cohorts, and examined associations with cognition and imaging outcomes, including synaptic vesicle glycoprotein 2A (SVA) positron emission tomography (PET). Methods SNAP25 was quantified by single-molecule array (SIMOA) in the BioFINDER-2 pilot (n = 102), Genova Memory Center (GMC; n = 106, plasma subset n = 42), and TRIAD (n = 36) cohorts. Participants were cognitively unimpaired (CU) or impaired (CI) and further stratified based on amyloid status (+/-). Group differences were tested adjusting for age, sex, and education, with Tukey corrections and diagnostic performance by ROC analyses and DeLong tests. Plasma-CSF effects were compared using mixed-effects models and Wald tests, and associations by Spearman/partial Spearman correlations and linear regressions. Plasma neurofilament light was used as an SV2A PET comparator. Results Plasma SNAP25 levels were elevated in CI+ across all cohorts (fold change = 1.46 in BioFINDER-2, 1.48 in GMC, and 1.77 in TRIAD, p < 0.01) compared to CU−, with AUCs of 0.78, 0.92, and 0.90 respectively. CSF group differences were less consistent, as were plasma-CSF correlations of SNAP25. Furthermore, in the BioFINDER-2, higher SNAP25 was associated with lower Mini-Mental State Examination scores, more brain atrophy, and greater amyloid and tau PET burden. Amyloid PET associations were also observed with CSF SNAP25 in GMC, and with plasma SNAP25 in TRIAD. In the full TRIAD cohort, plasma SNAP25 correlated with global SV2A PET meta-ROI after adjustment for age and sex (r s =0.36, p = 0.04). In CI+ participants, there was also a significant association with global SV2A PET binding. Voxelwise plasma associations were strongest in the ventromedial prefrontal cortex. NfL was not associated with SV2A PET binding. Conclusion Plasma SNAP25 is reliably measurable and consistently elevated in CI+ individuals. Associations with core AD biomarkers and SV2A PET binding, particularly in CI+ individuals, support its potential as a marker of synaptic injury for disease characterisation and clinical-trial monitoring.