Integrative analysis of cerebrospinal fluid biomarkers, metabolomics, and polygenic risk reveals novel metabolite associations with Alzheimer's disease

Background: Cerebrospinal fluid (CSF) metabolomics offers an opportunity to investigate in vivo biological pathways impacted in the human brain by Alzheimer's disease (AD). While impairments in brain glucose metabolism and lipid homeostasis are implicated in AD, the underlying metabolic pathways remain unclear. Genotype information can also be leveraged to study associations between CSF metabolites and AD genetic risk. Objective: To evaluate how CSF metabolomic profiles and genetic risk are associated with AD pathology as reflected by established CSF biomarkers (Aβ, P-Tau, and T-Tau). Methods: We collected CSF mass spectrometry measurements of 678 metabolites and 4865 unnamed compounds, as well as genome-wide genotype data from 487 individuals in the Amsterdam Dementia cohort. Polygenic risk scores (PRS) for AD were calculated. Elastic net regression models were used to predict AD biomarker levels with CSF metabolites, and pathway enrichment analysis was performed to assess the metabolic pathways involved. Results: 98 CSF metabolites were found to be significantly correlated with P-Tau or T-Tau, but none with Aβ CSF levels. Elastic net regression models identified 42 and 34 metabolites predicting P-Tau and T-Tau, respectively, including novel associations with Anserine and Fucose. Pathway enrichment analysis implicated Pentose and Glucuronate Interconversions, Glycerophospholipid Metabolism, and ABC Transporters in AD pathology. PRS analysis highlighted four CSF phosphatidylcholines significantly associated with AD genetic risk. Conclusions: CSF metabolites demonstrate a lack of Aβ levels associations, contrasting with multiple significant findings for P-Tau and T-Tau. Novel associations with Anserine and Fucose may provide new insights into metabolic pathways impacted by AD pathology.