Serum β-synuclein predicts dementia conversion and mortality beyond amyloid-β and phospho-tau

Synaptic dysfunction is among the earliest pathological events in Alzheimer's disease and among the strongest neuropathological correlates of cognitive impairment, yet blood-based synaptic biomarkers are lacking. Whether serum β-synuclein provides prognostic information independent of established amyloid-β (A; CSF Aβ42/Aβ40 ratio) and tau (T1; CSF pTau181) biomarkers has not been determined. We studied 259 non-demented individuals with subjective or mild cognitive impairment from a single-centre memory clinic, classified by CSF biomarkers using externally validated cut-offs into A-T1- controls (n = 97), A-T1+ PASSED ('pTau and Aβ surge with subtle deterioration', n = 65), and A + T1+ Alzheimer's disease (n = 97), with parallel analyses using Aβ40-normalised pTau181. Endpoints comprised dementia conversion (n = 185, median follow-up 4.1 years), all-cause mortality (n = 258, median 7.0 years), longitudinal MMSE trajectories (n = 185), and regional brain volumes (n = 137). Independence from CSF biomarkers was quantified through systematic effect attenuation analysis. Serum β-synuclein was selectively elevated in the Alzheimer's disease group (Cohen's d = 0.69 vs controls, P < 0.001) but not in PASSED individuals (P = 0.40), and was more strongly associated with CSF pTau181 (r = 0.36) than with the Aβ42/Aβ40 ratio (r = -0.20) across the cohort. Each standard deviation increase was associated with higher risk of dementia conversion (hazard ratio (HR) = 1.40; 95% CI: 1.11, 1.77; P = 0.005), faster cognitive decline (-0.11 SD/year, 95% CI: -0.18, -0.04, P = 0.002, corresponding to -0.20 MMSE points/year), and higher mortality (HR = 1.33; 95% CI: 1.07, 1.65; P = 0.010). For dementia conversion and mortality, adjustment for continuous CSF pTau181 and Aβ42/Aβ40 produced minimal attenuation (≤13%). Both associations were preserved under Aβ40-normalised T1 staging but not against continuous pTau181 normalised to CSF Aβ40. For cognitive decline, β-synuclein's effect remained significant after Aβ42/Aβ40 adjustment (25% attenuation; P = 0.007) but was substantially attenuated by pTau181 (58.5%; P = 0.186). Serum β-synuclein was associated with lower bilateral temporal lobe volume-the only lobar regions to survive hierarchical testing-with sub-regional involvement spanning the hippocampi, entorhinal cortex and parahippocampal gyrus, and explained 5.0-8.9% of unique variance beyond CSF biomarkers. Plasma glial fibrillary acidic protein (GFAP) showed a strong association with β-synuclein, estimated in the PASSED subgroup (β=0.34 per SD, P < 0.001). Serum β-synuclein provided prognostic information largely independent of CSF Aβ42/Aβ40 across all endpoints and of pTau181 for survival outcomes, while sharing substantial prognostic variance with pTau181 for cognitive trajectory, supporting its candidacy as a blood-based synaptic complement to core Alzheimer's disease biomarkers.