Best combination of CSF biomarkers for predicting cognitive decline and clinical progression: A multi‐cohort study. (20th December 2022)

• Record Type: Journal Article
• Authors: Salvadó, Gemma; Larsson, Victoria; Cody, Karly Alex; Cullen, Nicholas; Jonaitis, Erin M.; Stomrud, Erik; Kollmorgen, Gwendlyn; Wild, Norbert; Palmqvist, Sebastian; Janelidze, Shorena; Mattsson‐Carlgren, Niklas; Zetterberg, Henrik; Blennow, Kaj; Johnson, Sterling C.; Ossenkoppele, Rik; Hansson, Oskar
• Is Part Of: Alzheimer's & dementia. Volume 18(2022)Supplement 6
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-12-20
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.066480
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Amyloid‐β (Aβ) and tau markers have very good prognostic value for Alzheimer's disease (AD); however, novel neurodegeneration and glial‐related cerebrospinal fluid (CSF) biomarkers may help further improve prognosis. Using multiple CSF biomarkers from a single multi‐panel, we explored the combination of biomarkers that best predicts cognitive decline and clinical progression in three separate cohorts with relevance for clinical practice and trials. Method: In this study, we included 1, 991 participants (cognitively unimpaired [CU], n=1, 462; mild cognitive impairment [MCI], n=529) from three separate cohorts (BioFINDER‐1, BioFINDER‐2 and Wisconsin's: Wisconsin Registry for Alzheimer's Prevention [WRAP] & Wisconsin Alzheimer's Disease Research Center [WADRC]; Table 1) with longitudinal cognitive and clinical measurements. CSF was measured in all participants using the NeuroToolKit panel (a panel of automated Elecsys® and robust prototype immunoassays, Roche Diagnostics International Ltd) for biomarkers of AD pathology (p‐tau/Aβ42 ratio), neurodegeneration (neurofilament light chain [NfL], neurogranin, and α‐synuclein), and glial function (sTREM2, GFAP, YKL‐40, IL‐6, and S100b). Linear mixed models and generalized linear mixed models were used in each cohort to identify biomarker combinations most predictive of clinical progression (to Alzheimer's dementia or all‐cause dementia) or cognitive decline (Preclinical Alzheimer's cognitive composite [PACC] for CU and Mini‐Mental State Examination [MMSE] for MCI). Best performing models were compared to single‐predictor (p‐tau/Aβ42) models using Akaike information criterion (AIC) and areas under the curve (AUC), respectively. Result: Only p‐tau/Aβ42 ratio was selected as predictor in the parsimonious models for predicting conversion to Alzheimer's dementia in all cohorts and conversion to all‐cause dementia in the Wisconsin's cohort (Table 2, Figure 1). In the other two cohorts, model performance was significantly improved by the addition of NfL and neurogranin (BioFINDER‐1) and NfL alone (BioFINDER‐2). For predicting cognitive decline, parsimonious models included p‐tau/Aβ42 ratio alone (MMSE in BioFINDER‐1 and PACC in the other cohorts) or together with NfL (rest) as predictors. Conclusion: The CSF p‐tau/Aβ42 ratio is sufficient for prediction of AD progression, showing very high accuracy in three separate cohorts. The addition of NfL to p‐tau/Aβ42 can be applied to a clinical setting to further improve the prediction of all‐cause dementia and cognitive decline.