Optimal combinations of CSF biomarkers for predicting cognitive decline and clinical conversion in cognitively unimpaired participants and mild cognitive impairment patients: A multi-cohort study

Introduction: Our objective was determining the optimal combinations of cerebrospinal fluid (CSF) biomarkers for predicting disease progression in Alzheimer's disease (AD) and other neurodegenerative diseases. Methods: We included 1,983 participants from three different cohorts with longitudinal cognitive and clinical data, and baseline CSF levels of Aβ42, Aβ40, phosphorylated tau at threonine-181 (p-tau), neurofilament light (NfL), neurogranin, α-synuclein, soluble triggering receptor expressed on myeloid cells 2 (sTREM2), glial fibrillary acidic protein (GFAP), YKL-40, S100b, and interleukin 6 (IL-6) (Elecsys NeuroToolKit). Results: Change of modified Preclinical Alzheimer's Cognitive Composite (mPACC) in cognitively unimpaired (CU) was best predicted by p-tau/Aβ42 alone (R ² ≥ 0.31) or together with NfL (R ² = 0.25), while p-tau/Aβ42 (R ² ≥ 0.19) was sufficient to accurately predict change of the Mini-Mental State Examination (MMSE) in mild cognitive impairment (MCI) patients. P-tau/Aβ42 (AUC ≥ 0.87) and p-tau/Aβ42 together with NfL (AUC ≥ 0.75) were the best predictors of conversion to AD and all-cause dementia, respectively. Discussion: P-tau/Aβ42 is sufficient for predicting progression in AD, with very high accuracy. Adding NfL improves the prediction of all-cause dementia conversion and cognitive decline.