Fig. 2: BAs, BCAAs, and excitatory neurotransmitters exhibit strong associations with AD.
From: Metabolic phenotyping reveals an emerging role of ammonia abnormality in Alzheimer’s disease
a Composition of metabolic profiles stratified by metabolite type. b The partial least squares discriminant analysis (PLS-DA) scores plot generated by the raw concentrations of all metabolites and samples. The circles and error bars represent the centroids (mean+S.E.) of principal component scores corresponding to each stage. c Fold changes of PC1s (the first component of PCA) derived from the metabolite types in subjects with SCD, MCI, and AD relative to CN. * indicates ANOVA FDR < 0.05 (two-sided) when comparing NC, SCD, MCI, and AD. $ indicates post hoc Dunnett’s test p < 0.05 (two-sided) when compared to CN. Metabolite types are arranged in decreasing order of ANOVA FDR values. The number next to the name represents the percentage of variation that PC1 captured. d Levels (mean with S.E.) of five sub-types belonging to the top three types of (c) in four clinical stages. The levels are represented by the PC1 scores derived from PCA based on metabolites belonging to each sub-type and were scaled to the same starting point (n = 487, 239, 284, 387 for CN, SCD, MCI, and AD respectively). The number next to the name represents the percentage of variation that PC1 captured. Source data are provided as a Source Data file.
