Antibody-based methods for the measurement of α-synuclein concentration in human cerebrospinal fluid - method comparison and round robin study

Abstract

α-Synuclein is the major component of Lewy bodies and a candidate biomarker for neurodegenerative diseases in which Lewy bodies are common, including Parkinson's disease and dementia with Lewy bodies. A large body of literature suggests that these disorders are characterized by reduced concentrations of α-synuclein in cerebrospinal fluid (CSF), with overlapping concentrations compared to healthy controls and variability across studies. Several reasons can account for this variability, including technical ones, such as inter-assay and inter-laboratory variation (reproducibility). We compared four immunochemical methods for the quantification of α-synuclein concentration in 50 unique CSF samples. All methods were designed to capture most of the existing α-synuclein forms in CSF ('total' α-synuclein). Each of the four methods showed high analytical precision, excellent correlation between laboratories (R² 0.83-0.99), and good correlation with each other (R² 0.64-0.93), although the slopes of the regression lines were different between the four immunoassays. The use of common reference CSF samples decreased the differences in α-synuclein concentration between detection methods and technologies. Pilot data on an immunoprecipitation mass spectrometry (IP-MS) method is also presented. Our results suggest that the four immunochemical methods and the IP-MS method measure similar forms of α-synuclein and that a common reference material would allow harmonization of results between immunoassays.

Keywords: biomarker; cerebrospinal fluid; enzyme-linked immunoabsorbent assay; mass spectrometry; round robin; α-synuclein.

Figure 1

Scatter plots for MSD (a), BioLegend (b), ADx (c), and Roche (d) CSF α‐synuclein assays. On the y‐axis of each panel, each sample's concentration per site (i.e., averaged across replicates and kit lots) is shown. The corresponding average concentration across all sites is given on the x axis. The results from linear regression of each site's concentrations against the average are shown, with corresponding number of the 49 samples, slope, and coefficient of determination (R ²). The dotted line indicates x = y.

Figure 2

Variance components (expressed as % CV) per assay. Variance component analysis partitions the total variance of all assays associated with each sample's measurements (n = 49) into site components, kit lot, and replicate. The distribution of the components by assay are highlighted by the box and violin plots. Individual samples are joined by the light gray lines; no one sample appears to have elevated variance components across assays. The impact of Goettingen's unusual results in the BioLegend assay has been removed from the analysis, leaving the BioLegend assay with variance components comparable to the other assays. (The total and site variance components with Goettingen included are shown by the dotted lines.) Note that Roche measured their samples with the equivalent of just one kit lot; hence, that assay has no kit lot variance component.

Figure 3

Relationship between the CSF α‐synuclein concentrations for MSD in y against BioLegend (a), ADx (b), and Roche (c) in x; for BioLegend in y against MSD (d), ADx (e), and Roche (f) in x; for ADx in y against MSD (g), BioLegend (h), and Roche (I) in x; and Roche in y against MSD (j), BioLegend (k), and ADx (l) in x. All concentrations of 49 samples measured at the originating laboratories (the laboratories that developed the respective assay) and have been averaged across replicates and kit lots.

Figure 4

Association between average α‐synuclein concentrations determined by five N‐Terminal fragments and three C‐Terminal fragments from the IP‐MS mass spectrometry procedure. The dotted line indicates the identity y=x. Passing‐Bablok regression results are shown, with the estimate (solid line) and corresponding 95% confidence intervals (curved dashes).