Clinical validation of the Lumipulse G cerebrospinal fluid assays for routine diagnosis of Alzheimer’s disease

Maria João Leitão,Anuschka Silva-Spínola,Isabel Santana,Veronica Olmedo,Alicia Nadal,Nathalie Le Bastard,Inês Baldeiras

DOI: https://doi.org/10.1186/s13195-019-0550-8

2019-11-23

Alzheimer's Research and Therapy

Abstract:Abstract Background Ongoing efforts within the Alzheimer’s disease (AD) field have focused on improving the intra- and inter-laboratory variability for cerebrospinal fluid (CSF) biomarkers. Fully automated assays offer the possibility to eliminate sample manipulation steps and are expected to contribute to this improvement. Recently, fully automated chemiluminescence enzyme immunoassays for the quantification of all four AD biomarkers in CSF became available. The aims of this study were to (i) evaluate the analytical performance of the Lumipulse G β-Amyloid 1-42 (restandardized to Certified Reference Materials), β-Amyloid 1-40, total Tau, and pTau 181 assays on the fully automated LUMIPULSE G600II; (ii) compare CSF biomarker results of the Lumipulse G assays with the established manual ELISA assays (INNOTEST®) from the same company (Fujirebio); and (iii) establish cut-off values and the clinical performance of the Lumipulse G assays for AD diagnosis. Methods Intra- and inter-assay variation was assessed in CSF samples with low, medium, and high concentrations of each parameter. Method comparison and clinical evaluation were performed on 40 neurological controls (NC) and 80 patients with a diagnosis of probable AD supported by a follow-up ≥ 3 years and/or positive amyloid PET imaging. A small validation cohort of 10 NC and 20 AD patients was also included to validate the cut-off values obtained on the training cohort. Results The maximal observed intra-assay and inter-assay coefficients of variation (CVs) were 3.25% and 5.50%, respectively. Method comparisons revealed correlation coefficients ranging from 0.89 (for Aβ40) to 0.98 (for t-Tau), with those for Aβ42 (0.93) and p-Tau (0.94) in-between. ROC curve analysis showed area under the curve values consistently above 0.85 for individual biomarkers other than Aβ40, and with the Aβ42/40, Aβ42/t-Tau, and Aβ42/p-Tau ratios outperforming Aβ42. Validation of the cut-off values in the independent cohort showed a sensitivity ranging from 75 to 95% and a specificity of 100%. The overall percentage of agreement between Lumipulse and INNOTEST was very high (> 87.5%). Conclusions The Lumipulse G assays show a very good analytical performance that makes them well-suited for CSF clinical routine measurements. The good clinical concordance between the Lumipulse G and INNOTEST assays facilitates the implementation of the new method in routine practice.