Multiplex proteomics identifies novel CSF and plasma biomarkers of early Alzheimer's disease

Abstract

To date, the development of disease-modifying therapies for Alzheimer's disease (AD) has largely focused on the removal of amyloid beta Aβ fragments from the CNS. Proteomic profiling of patient fluids may help identify novel therapeutic targets and biomarkers associated with AD pathology. Here, we applied the Olink™ ProSeek immunoassay to measure 270 CSF and plasma proteins across 415 Aβ- negative cognitively normal individuals (Aβ- CN), 142 Aβ-positive CN (Aβ+ CN), 50 Aβ- mild cognitive impairment (MCI) patients, 75 Aβ+ MCI patients, and 161 Aβ+ AD patients from the Swedish BioFINDER study. A validation cohort included 59 Aβ- CN, 23 Aβ- + CN, 44 Aβ- MCI and 53 Aβ+ MCI. To compare protein concentrations in patients versus controls, we applied multiple linear regressions adjusting for age, gender, medications, smoking and mean subject-level protein concentration, and corrected findings for false discovery rate (FDR, q < 0.05). We identified, and replicated, altered levels of ten CSF proteins in Aβ+ individuals, including CHIT1, SMOC2, MMP-10, LDLR, CD200, EIF4EBP1, ALCAM, RGMB, tPA and STAMBP (- 0.14 < d < 1.16; q < 0.05). We also identified and replicated alterations of six plasma proteins in Aβ+ individuals OSM, MMP-9, HAGH, CD200, AXIN1, and uPA (- 0.77 < d < 1.28; q < 0.05). Multiple analytes associated with cognitive performance and cortical thickness (q < 0.05). Plasma biomarkers could distinguish AD dementia (AUC = 0.94, 95% CI = 0.87-0.98) and prodromal AD (AUC = 0.78, 95% CI = 0.68-0.87) from CN. These findings reemphasize the contributions of immune markers, phospholipids, angiogenic proteins and other biomarkers downstream of, and potentially orthogonal to, Aβ- and tau in AD, and identify candidate biomarkers for earlier detection of neurodegeneration.

Keywords: Alzheimer’s disease; Angiogenesis; Apoptosis; Biomarker; Inflammation; Mild cognitive impairment; Proteomics.

Fig. 1

Proteins showing evidence of differential regulation in two or more Aβ+ groups, in a cerebrospinal fluid (CSF) and b plasma. Dashed line represents the average protein concentration of the Aβ- cognitively normal group (i.e. study controls). Aβ- groups are coloured in blue; Aβ+ groups are coloured in red. Ctrl-ABneg = Aβ- controls; MCI-ABneg = Aβ- MCI patients; Ctrl-ABpos = Aβ+ cognitively normal individuals, MCI- ABpos = Aβ+ MCI patients; AD = Dementia due to Alzheimer’s disease; q = false discovery rate adjusted p-value. Image generated using the ggplot package in R

Fig. 2

Volcano plots illustrating the log₂-transformed fold change and -log₁₀-transformed p-value (uncorrected) for all proteins assessed in a cerebrospinal fluid and b plasma. Proteins showing evidence of differential regulation after adjustment for false discovery rate (FDR) are denoted in blue. Proteins showing evidence of differential regulation in two or more patient groups, and/or showing replicated evidence of differential regulation in the Memory Lund replication sample, are noted in green text

Fig. 3

Between- and within-tissue correlation matrix, representing the strength of correlation (Pearson’s r) between proteins in plasma and cerebrospinal fluid (CSF). Positive correlations are illustrated in blue; negative correlations are illustrated in red. To reduce the dimensionality of the correlation matrix, only those proteins that were above the limit of detection, yielding Pearson’s r values equal to or greater than 0.7, were included. Image generated using the ggplot package in R