AUTHOR=Oudart Jean-Baptiste , Djerada Zoubir , Nonnonhou Vignon , Badr Sarah , Bertholon Laurie-Anne , Dammak Anis , Jaidi Yacine , Novella Jean-Luc , Pallet Nicolas , Gillery Philippe , Mahmoudi Rachid TITLE=Incremental Value of CSF Biomarkers in Clinically Diagnosed AD and Non-AD Dementia JOURNAL=Frontiers in Neurology VOLUME=Volume 11 - 2020 YEAR=2020 URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2020.00560 DOI=10.3389/fneur.2020.00560 ISSN=1664-2295 ABSTRACT=Background: Cerebrospinal fluid (CSF) biomarkers are used to diagnose Alzheimer disease (AD), especially in atypical clinical presentations. No consensus currently exists regardingcut-off values. This study aimed, first, to define optimal cut-off values for CSF biomarkers, and second, to investigate the added value of different classification strategies based on CSF biomarkers. Methods: A total of 380 patients were prospectively included in this study: 140 with AD, and 240 with various diagnoses (non-AD). CSF biomarkers were measured using ELISA. Six classification strategies were compared based on different combinations of T-Tau, P-Tau181, Aβ42, Aβ40assays and Aβ42/Aβ40 ratiocalculation. Results: ROC curve analysis of T-Tau, P-Tau181, Aβ42, Aβ40concentrations and Aβ42/Aβ40 ratio levels showed AD cut-off values of ≥355 ng/mL, ≥57 ng/mL, ≤706 ng/mL, ≥10,854 ng/mL and ≤0.059 respectively. The highest correct classificationrates were found with the two-step algorithm associating the standard classification strategy (i.e. T-Tau, P-Tau181, Aβ42 concentrations)with Aβ40 concentration cut-off in case of conflicting results (71.6%). However, this method had also the highest misclassification rates, which makes it unsuitable in daily practice due to lack of specificity. Compared to this algorithm, the addition of the Aβ42/Aβ40ratioin case of conflicting results showed good correct classification (65.3%) and a lower misclassification rates (12.6% vs 22.1%, respectively). Conclusion: This study showed that the two-step algorithm associating standard classification strategy with Aβ42/Aβ40ratio cut-off in case of conflicting results, appears to be the most efficient strategy among those tested in this study.