Combination of CSF synaptic biomarkers and qEEG leads to a better prediction of disease progression in Alzheimer’s disease

Abstract Background Synaptic dysfunction plays an important role in Alzheimer’s disease (AD) and cognitive decline. We investigated the association between cerebrospinal fluid (CSF) synaptic protein levels and quantitative EEG (qEEG) measures, two modalities to measure synaptic dysfunction in AD pathology. We assessed combined and independent prognostic value of both modalities for cognitive decline along the AD continuum. Method We included 114 subjects with SCD or MCI (n=38 with normal Aβ42 [A‐], n=38 with abnormal CSF Aβ42 [A+], and 38 patients with AD dementia) from the Amsterdam dementia cohort, with available CSF and EEG at baseline, and more than one year clinical follow‐up. We assessed associations between CSF synaptic proteins (neurogranin, β‐synuclein, neuronal pentraxin 2 [NPTX2], synaptotagmin‐1 [SYT1], SNAP25) and qEEG measures (Global field power [GFP] ‐α,‐β,‐θ; Amplitude envelope correlation‐corrected [AEC‐c] α, AEC‐c β, and phase lag index [PLI] θ), all standardized into z‐scores, using Spearman correlations. We performed linear mixed models to assess univariate and multivariate associations with cognitive decline (i.e. repeated MMSE) along the AD continuum. We used backwards stepwise selection in multivariate models and calculated marginal R‐squared to compare prognostic accuracy. Result There were no correlations between CSF and qEEG markers (Table 1). In univariate models SNAP25, beta‐synuclein, neurogranin, GFP alpha, beta and theta emerged as predictors of cognitive decline (Table 2). SNAP25 showed the largest effect (stB[SE]=‐0.45 [0.13], p<0.001). For GFP‐alpha and –beta effects were reversed (0.29[0.13], p<0.05 and 0.37[0.13], p<0.01 respectively). Multivariate models combining CSF and qEEG measures improved prognostic accuracy compared to CSF or EEG only models (Table 3; R 2 m = 19.8%, StB[SE] = 1.26[0.28] for GFP‐α, StB[SE] = 1.32[0.32] for GFP‐β, StB[SE] = ‐0.96[0.32] for SNAP25 and StB[SE] = 0.94[0.31] for NPTX2, all p<0.01). Conclusion CSF synaptic proteins and qEEG measures were not directly correlated to each other, but both modalities were associated to cognitive decline. This suggests they represent different aspects of synaptic pathology in AD. The combination of both modalities showed increased prognostic accuracy over CSF‐ or qEEG‐only models. Hence, the combination has potential for a more personalized prognosis in AD.