Longitudinal dynamics of cerebrospinal fluid Aꞵ, pTau and sTREM2 reveal predictive preclinical trajectories of Alzheimer’s pathology
Abstract
Alzheimer's disease (AD) is a prevalent neurodegenerative disorder with limited early-stage treatment options. There is an urgent and unmet need for accurate biomarkers which can identify patients at risk for AD before cognitive symptoms emerge. Here I compared the performance of two analytical methods, univariate and multivariate classification, for identifying cognitively normal (CN) and mild cognitive impairment (MCI) patients based on their cerebrospinal fluid (CSF) biomarkers of Aβ42, pTau-181, sTREM2. Post-hoc analyses were then employed to assess patient progression in each of the SNF clusters. I found that SNF identified subgroups within the CN and MCI cohorts, based solely on conjoint patterns of CSF, uncaptured by univariate strategies. In both CN and MCI, a fast progressor patient cluster was identified. Our findings suggest that multivariate modeling of CSF data can uncover predictive patterns of AD progression which may help to stratify patients in clinical trials of preventative therapeutics.