The cerebrospinal fluid biomarker ratio Aβ42/40 identifies amyloid positron emission tomography positivity better than Aβ42 alone in a heterogeneous memory clinic cohort

Abstract

Abstract Background Cerebrospinal fluid (CSF) analysis for detecting amyloid positivity may be as reliable as positron emission tomography (PET). We evaluated the performance of the amyloid beta (Aβ)42/40 ratio for predicting amyloid positivity by PET, compared with Aβ42 alone, and phosphorylated tau 181 (pTau181)/Aβ42 and total tau (tTau)/Aβ42 ratios, using fully automated CSF immunoassays (Roche Diagnostics International Ltd, Rotkreuz, Switzerland) in a heterogeneous cohort of patients with a range of cognitive disorders reflecting the typical population of a memory clinic. Methods CSF samples from 103 patients with known amyloid PET status (PET positive = 54; PET negative = 49) were retrospectively selected from one site in Germany; 71 patients were undergoing treatment for mild cognitive impairment (n = 44) or mild-to-moderate dementia (n = 27) due to Alzheimer’s disease (AD), and 32 patients were undergoing treatment for non-AD-related cognitive disorders. Aβ42, pTau181, and tTau concentrations were measured in CSF samples using the respective Elecsys® CSF immunoassays modified for use on the cobas e 411 analyzer; Aβ40 concentrations were measured using a non-commercially available robust prototype assay. Sensitivities/specificities for amyloid positivity cut-offs (Youden-derived and pre-defined) were calculated, and receiver operating characteristic analyses determined area under the curve (AUC) versus amyloid PET status. Limitations include a small sample size, use of a pre-analytical protocol not in accordance with the Elecsys CSF immunoassay method sheets, and the lack of a pre-defined cut-off for Aβ42/40. Results Point estimates for sensitivity and specificity of CSF biomarkers and biomarker ratios versus amyloid PET were 0.93 and 0.57 for Aβ42, 0.96 and 0.69 for pTau181/Aβ42, 0.92 and 0.69 for tTau/Aβ42, and 0.94 and 0.82 for Aβ42/40. For AUCs, point estimates (95% confidence intervals) versus amyloid PET were 0.78 (0.68−0.88) for Aβ42, 0.88 (0.81−0.95) for pTau181/Aβ42, 0.87 (0.80−0.95) for tTau/Aβ42, and 0.90 (0.83−0.97) for Aβ42/40. Conclusions CSF Aβ42/40 ratio can predict PET amyloid positivity with high accuracy in patients with a range of cognitive disorders when evaluating Aβ pathology independent of tau and neurodegeneration for research purposes. The performance of Aβ42/40 was comparable with pTau181/Aβ42 and tTau/Aβ42 used in clinical practice and better than Aβ42 alone.