Abstract
BACKGROUND: In this study, we tested to which extent possible between-center differences in standardized operating procedures (SOPs) for biobanking of cerebrospinal fluid (CSF) samples influence the homogeneity of the resulting aliquots and, consequently, the concentrations of the centrally analyzed selected Alzheimer's disease biomarkers. METHODS: Proficiency processing samples (PPSs), prepared by pooling of four individual CSF samples, were sent to 10 participating centers, which were asked to perform aliquoting of the PPSs into two secondary aliquots (SAs) under their local SOPs. The resulting SAs were shipped to the central laboratory, where the concentrations of amyloid beta (Aβ) 1-42, pTau181, and albumin were measured in one run with validated routine analytical methods. Total variability of the concentrations, and its within-center and between-center components, were analyzed with hierarchical regression models. RESULTS: We observed neglectable variability in the concentrations of pTau181 and albumin across the centers and the aliquots. In contrast, the variability of the Aβ1-42 concentrations was much larger (overall coefficient of variation 31%), with 28% of the between-laboratory component and 10% of the within-laboratory (i.e., between-aliquot) component. We identified duration of the preparation of the aliquots and the centrifugation force as two potential confounders influencing within-center variability and biomarker concentrations, respectively. CONCLUSIONS: Proficiency processing schemes provide objective evidence for the most critical preanalytical variables. Standardization of these variables may significantly enhance the quality of the collected biospecimens. Studies utilizing retrospective samples collected under different local SOPs need to consider such differences in the statistical evaluations of the data.