Longitudinal Dynamics of the Blood Proteome Within the Alzheimer Spectrum in Down Syndrome

BACKGROUND AND OBJECTIVES: Adults with Down syndrome (DS) have an increased risk of developing early Alzheimer disease. While our previous cross-sectional study investigated the blood proteome in the context of Alzheimer in DS, identifying CBLN4, CD14, C-X-C motif chemokine 17 (CXCL17), ectodysplasin A2 receptor (EDA2R), glial fibrillary acidic protein (GFAP), insulin-like growth factor binding protein-2 (IGFBP2), neurofilament light (NFL), SEPTIN3, and SPON1 as proteins of interest, longitudinal trajectories remain mostly unexplored. We aimed to characterize the longitudinal dynamics of these plasma proteins to evaluate their potential contribution to disease pathophysiology and associated cognitive decline. METHODS: Adults with diagnosis of DS and ability to understand instructions of neuropsychological assessments were recruited at Ludwig Maximilians University Hospital Munich, returning for at least 1 annual follow-up visit for repeated clinical assessment, neuropsychological testing, and blood sample collection after baseline visit. Longitudinal blood protein dynamics were assessed using OLINK technology. Bayesian modeling analyzed longitudinal protein trajectories and their impact on cognitive performance. Spearman correlations of regression-derived rates of change investigated relationships between proteins and cognitive subdomains stratified by clinical diagnosis. RESULTS: We included 59 adults with DS (46% female, median age = 32 years [interquartile range: 27-47]). All completed the first annual follow-up visit (median duration 13.37 months [12.35-20.95]), with 14 of them returning for a follow-up 2 visit (median 25.43 months [24.79-33.8]). Stratified analysis included 13 individuals with DS with (38% female, 54 years [49-58]) and 41 without diagnosis of cognitive decline (49% female, 29 years [24-35]). Cluster of differentiation 14, CXCL17, EDA2R, GFAP, IGFBP2, NFL, SEPTIN3, and SPON1 exhibited longitudinal increase, while CBLN4 decreased (all posterior distributions ≥ 99.12%), with age. Protein baseline levels were associated with cognitive decline over time while controlling for age in the whole cohort (all posterior distributions ≥ 94.75%). After correction for multiple comparisons, no longitudinal changes in protein markers exhibited a significant relationship when correlated with changes in cognitive subdomain performance within subcohorts. DISCUSSION: In DS, longitudinal characterization of all previously identified proteins revealed distinct trajectories over time and an association of baseline levels with future cognitive decline. This suggests potential contributions of these proteins to the complex underlying pathophysiology of Alzheimer in DS and should motivate further investigation.