CSF proteins of inflammation, proteolysis and lipid transport define preclinical AD and progression to AD dementia in cognitively unimpaired individuals

Abstract

This preclinical AD CSF proteome study identified a panel of 12-CSF markers detecting amyloid positivity and clinical progression to AD with high accuracy; some of these CSF proteins related to immune function, neurotrophic processes, energy metabolism and endolysosomal functioning (e.g., ITGB2, CLEC5A, IGFBP-1, CST3) changed before amyloid positivity is established.

Fig. 1

CSF proteins of inflammation, proteolysis and lipid transport define preclinical AD and predict progression to AD dementia in cognitively unimpaired individuals a Protein levels in CSF from cognitively unimpaired individuals with and without amyloid pathology (based on CSF Aß42 concentration; CU(Aß +) = 65; CU(Aß-) = 232) were measured using antibody-based PEA technology. We defined which proteins differed across groups and a classification model to identify preclinical AD cases, which was validated in an independent cohort of cognitively unimpaired volunteers (EMIF-AD cohort, CU(Aß-):103; CU(Aß +): 19). The performance of the model to predict clinical progression to cognitive impairment was also evaluated in a subset of cases from the discovery cohort (progressors to symptomatic stages: 39). We also evaluated the association of the proteins within the panel with proxies of progression of AD pathophysiology (e.g., CSF Aß42). b Volcano plot shows the CSF proteins that are differentially regulated in CU(Aß +) vs. CU(Aß-). Each dot represents a protein. The beta coefficients (log2 fold-change) are plotted versus q values (-log10-transformed FDR corrected p-value). Proteins significantly dysregulated after adjusting for false discovery rate (FDR, q < 0.05) are coloured in light green. The name of the top ten significant dysregulated CSF proteins and the top five with the strongest effect sizes are annotated. The total number of proteins that are down-regulated (left) or up-regulated (right) in CU(Aß +) is indicated. Horizontal dotted line indicates the significance threshold. Adjusted p values (q < 0.05) were calculated using a two-sided nested linear model adjusting for FDR. c UpSet plot indicates the overlap across the proteins regulated in the preclinical phase of AD (CU(Aß +) vs. CU(Aß-)) and those at the prodromal or dementia AD stage (MCI(Aß +) or AD (positive AD CSF profile) vs. controls) based on the results from our previous study [2]. d Receiver operating characteristic (ROC) curves depict the performance of 12-CSF protein panel to discriminate amyloid positive from amyloid negative cognitively unimpaired individuals in the ADC discovery and the EMIF-AD validation cohorts. In the discovery ADC cohort, black line is the mean Area Under the Curve (AUC) over all re-samplings (1000 repeats of fivefold cross-validation, grey lines). Inserts outline corresponding AUC and 95% CI. In the EMIF-AD validation cohort, insert outline the resulting AUC after directly applying the model developed with the discovery cohort. e ROC analysis depicts the performance of the CSF preAD panel to predict cases that progressed to MCI or dementia stage (progressors: 39; non progressors: 258). f CSF proteins within the preAD panel modelled along CSF Aβ42 as an early proxy of AD pathology progression. Each NPX protein value was transformed to z-scores based on the distribution in the actual dataset to allow visual comparison across proteins. Bold line indicates the mean trajectory and shadows the 95% CI. Dotted line depicts CSF Aß42 positivity threshold (< 813 pg/mL). CN, cognitively unimpaired controls; preAD: preclinical AD (CU individuals with amyloid pathology); MCI, Mild cognitive impairment; AD, Alzheimer’s disease; MMSE, Mini Mental Score Examination. Some images within Fig. 1a are courtesy of Olink® Proteomics AB