Biomarkers.

Background: Alzheimer's disease (AD) is identified by the accumulation of amyloid β (Aβ) and tau proteins in the brain. The NeuroToolKit offers automated cerebrospinal fluid (CSF) immunoassays of core AD biomarkers and biomarkers of neurodegeneration and synaptic function, including neurofilament light (NfL), SNAP-25, and neuronal pentraxin 2 (NPTX2). This work explores whether these three markers predict pre-dementia cognitive decline synergistically with or after accounting for CSF ptau/Aβ.

Method: Participants included N=360 baseline dementia-free adults (45-85 years) with complete CSF biomarker and cognitive data from the Wisconsin Registry for Alzheimer's Prevention and the Wisconsin Alzheimer's Disease Research Center (on-going, longitudinal preclinical AD studies). We assessed CSF biomarkers using the NeuroToolKit (NTK), a panel of robust prototype biomarker assays (Roche Diagnostics International, Switzerland). Mixed models examined separately whether NfL, SNAP-25, or NPTX2 biomarkers further modified Preclinical Alzheimer's Cognitive Composite (PACC3) trajectories beyond Aβ+ effects. Biomarker*ptau/Aβ*age (linear and quadratic) interactions were examined, and non-significant higher order interactions removed sequentially before interpreting NfL, SNAP-25, and NPTX2 effects.

Result: Sample characteristics are shown in Table 1, and show significant correlations among all biomarker pair except NPTX2 and ptau/Aβ. Table 2 and Figure 1 depict PACC3 model outputs and biomarker associations; Aβ+ performed worse or declined faster than Aβ- across all models. Compared to a base "amyloid-only" model, NfL, SNAP-25, and NPTX2 "biomarker" models showed an improved model fit (Table 2). The best-fitting model included the NPTX2* ptau/Aβ*age interaction (Table 2) which showed greatest average PACC3 decline in those with higher amyloid and lower NPTX2 (Figure 1d). SNAP-25 also related to cognition such that individuals with lower amyloid and higher SNAP-25 had the highest PACC3 average (interactions with age were non-significant), and individuals with higher NfL showed steeper PACC3 decline than those with lower NfL.

Conclusion: In addition to known effects of NfL, recently developed NTK immunoassays for SNAP-25 and NPTX2 accounted for additional variability on AD-related pre-dementia PACC3 performance. Future work may identify clinically useful cutpoints for SNAP-25 and NPTX2 and explore relations with other biomarkers common to neurodegeneration and AD. We will also examine associations with other outcomes including clinical progression of AD.