Mapping associations across multiple aspects of Alzheimer disease and the role of CSF biomarkers in individuals without dementia

Abstract

AbstractBackgroundAlzheimer disease (AD) is a complex disorder involving biological, clinical, and lifestyle aspects. Assessing all these aspects simultaneously allows to better disentangle the position of CSF marker abnormalities across AD stages. Here, we employ a novel statistical method specifically developed to reveal meaningful associations across multiple aspects that can occur simultaneously within individuals.MethodWe selected the baseline data from the European Prevention of Alzheimer Disease cohort of individuals without dementia (www.ep‐ad.org; n=1243, mean±SD age = 65±7 years old, 57% female, 18% CDR ≥0.5) covering ten aspects: demographics including APOE‐genotype, cerebrospinal fluid (CSF) biomarkers, MRI visual rating scales, MRI regional volumes, MRI white matter lesions, cognition, function, psychological factors, general physical health, and comorbidities. We used regularized simultaneous component analysis (regSCA), a statistical technique that extracts relevant components which explain variation within and across these aspects, and robustly selects the appropriate variables [1, 2].ResultThe optimal model revealed eight components explaining 51% of the total variance. Figure 1 presents an overview of all components, showing the strengths and directions of the associations between variables. Five components included CSF biomarkers, which were (1) Abeta42‐dominant, (2) Tau‐dominant, (3) function‐dominant, (4) age‐dominant and, (5) white matter lesion‐dominant. In (3) and (4), Abeta42 and Tau related to cognition and function. Only (3) was independent of age, and may represent purely AD. In (5), Abeta42 related to white matter lesions, and may represent the degree cerebral amyloid angiopathy (CAA). Abeta42 in (1) and Tau in (2) related to APOE‐genotype, suggesting these may be independent effects. The three components unrelated to CSF biomarkers were: (6) sex‐dominant, (7) psychological factors‐dominant, and (8) life‐style‐only.ConclusionAnalysis of robust associations across multiple aspects of AD simultaneously implicated CSF biomarkers in five components. These components may represent risk‐profiles and AD disease processes in cognitively normal and mildly impaired individuals. Our next step is to investigate the stability of the components and their co‐expression within individuals. This overview of complex relationships between AD‐related aspects, helps to understand the disease mechanisms in of AD. This novel approach can be a solution for analyses of large and complex datasets in AD.