Robustness of CSF Aβ42/40 and Aβ42/P‐tau181 measured using fully automated immunoassays to detect AD‐related outcomes

Abstract

IntroductionThis study investigated the comparability of cerebrospinal fluid (CSF) cutoffs for Elecsys immunoassays for amyloid beta (A beta)42/A beta 40 or A beta 42/phosphorylated tau (p-tau)181 and the effects of measurement variability when predicting Alzheimer's disease (AD)-related outcomes (i.e., A beta-positron emission tomography [PET] visual read and AD neuropathology). MethodsWe studied 750 participants (BioFINDER study, Alzheimer's Disease Neuroimaging Initiative [ADNI], and University of California San Francisco [UCSF]). Youden's index was used to identify cutoffs and to calculate accuracy (A beta-PET visual read as outcome). Using longitudinal variability in A beta-negative controls, we identified a gray zone around cut-points where the risk of an inconsistent predicted outcome was >5%. ResultsFor A beta 42/A beta 40, cutoffs across cohorts were <0.059 (BioFINDER), <0.057 (ADNI), and <0.058 (UCSF). For A beta 42/p-tau181, cutoffs were <41.90 (BioFINDER), <39.20 (ADNI), and <46.02 (UCSF). Accuracy was approximate to 90% for both A beta 42/A beta 40 and A beta 42/p-tau181 using these cutoffs. Using A beta-PET as an outcome, 8.7% of participants fell within a gray zone interval for A beta 42/A beta 40, compared to 4.5% for A beta 42/p-tau181. Similar findings were observed using a measure of overall AD neuropathologic change (7.7% vs. 3.3%). In a subset with CSF and plasma A beta 42/40, the number of individuals within the gray zone was approximate to 1.5 to 3 times greater when using plasma A beta 42/40. DiscussionCSF A beta 42/p-tau181 was more robust to the effects of measurement variability, suggesting that it may be the preferred Elecsys-based measure in clinical practice and trials.