Discriminatory and predictive capabilities of enzyme-linked immunosorbent assay and multiplex platforms in a longitudinal Alzheimer’s disease study

Abstract

BackgroundMultiplex assays such as xMAP have been proposed for the assessment of Alzheimer’s disease (AD) biomarkers amyloid β 42 (Aβ42), tau (Tau), and phosphorylated tau (pTau) in cerebrospinal fluid (CSF). Here, we compared the traditional enzyme-linked immunosorbent assay (ELISA) and xMAP with respect to their: (1) absolute biomarker concentration, (2) ability to distinguish AD from nondemented subjects, (3) ability to monitor AD longitudinally, and (4) ability to predict progression from mild cognitive impairment (MCI) to AD. MethodsWe selected 68 AD, 62 MCI, and 24 nondemented subjects, performed clinical examinations, and obtained CSF at baseline and 2 years later. Aβ42, Tau, and pTau were measured with both ELISA and xMAP. ResultsBiomarker levels differed considerably between the two assays, and the differences were concentration dependent. No differences were observed in ability to distinguish nondemented subjects from AD patients between ELISA (area under curve of 0.84 for Aβ42, 0.79 for Tau, and 0.75 for pTau) and xMAP (area under curve of 0.82 for Aβ42, 0.75 for Tau, and 0.73 for pTau), all P < .05. Increased Aβ42 levels of AD patients at follow-up compared with baseline were detected with ELISA, whereas increased Tau levels for nondemented subjects and MCI patients were only detected with xMAP. The hazard ratios for progression from MCI to AD did not differ between the assays. ConclusionBoth ELISA and multiplex assays can be used to measure AD biomarker levels in CSF to support clinical diagnosis and predict progression from MCI to AD with similar accuracy. Importantly, the assays’ output in absolute biomarker concentrations is remarkably different, and this discrepancy cannot be reconciled with simple correction factors.