Brain and blood metabolome for Alzheimer's dementia: findings from a targeted metabolomics analysis

Abstract

The development of Alzheimer's dementia (AD) accompanies both central and peripheral metabolic disturbance, but the metabolic basis underlying AD and metabolic markers predictive of AD risk remain to be determined. It is also unclear whether the metabolic changes in the peripheral blood and brain are overlapping in relation to AD. The present study addresses these questions by targeted metabolomics in both antemortem blood and postmortem brain samples in 2 community-based longitudinal cohorts of aging and dementia. We found that higher serum levels of 3 acylcarnitines, including decanoylcarnitine (C10), pimelylcarnitine (C7-DC), and tetradecadienylcarnitine (C14:2), significantly predict a lower risk of incident AD (composite hazard ratio = 0.368, 95% CI [0.207, 0.653]) after an average of 4.5-year follow-up, independent of age, sex, and education. In addition, baseline serum levels of ten glycerophospholipids, one amino acid, and 5 acylcarnitines predict the longitudinal change in cognitive functions. Moreover, 28 brain metabolites were associated with AD phenotypes. Of the putative metabolites identified in the serum and brain, 4 metabolites (3 glycerophospholipids [PC aa C30:0, PC ae C34:0, PC ae C36:1] and 1 acylcarnitine [C14:2]) were present in both the postmortem brain and antemortem blood, but only one metabolite (C14:2) was associated with AD in the same direction (i.e., protective). Partial correlation and network analyses suggest a potential tissue-specific regulation of metabolism, although other alternatives exist. Together, we identified significant associations of both central and peripheral metabolites with AD phenotypes, but there seems to be little overlap between the 2 tissues.

Keywords: AD neuropathology; Alzheimer's dementia; Cognitive decline; Postmortem brain metabolomics; Targeted metabolomics.

Figure 1,

Experimental design, sample size description, and statistical analysis strategy. Figure IA shows the experimental design, sample size description, and number of incident AD for the participants with blood samples; Figure IB shows the experimental design and sample size description for the participants with brain samples; and Figure 1C shows the overall statistical analysis strategy. For the analyses with serum or brain metabolites, outcome names were used in lieu of the association analysis with these outcomes.

Figure 2.

The Kaplan Meier survival curves of three serum acylcarnitine metabolites that are predictive of risk for incident AD (FDR ≤ 10%). Subjects (N = 530) were dichotomized based on the median level of metabolite concentrations. HR denotes the hazard ratio per fold metabolite level increase.

Figure 3.

Baseline serum level of PC ae C34:0 predicts cognitive decline over time (average 4.5-year follow-up). Shaded areas indicate standard errors for the trajectories. This analysis included 530 subjects who were NCI at baseline.

Figure 4.

Putative brain metabolites associated with measures for cognition and AD neuropathology (FDR≤10%). (A) Dichotomized by direction of associations; (B) Associations based on compound classes.

Figure 5.

Association of the symmetric dimethylarginine (SDMA) with various neuropathologic phenotypes for AD. LOR represents the log odds ratio of being in higher group per fold change of metabolite level, and beta represents the regression coefficient. This analysis included 111 subjects with brain samples.

Figure 6.

Summarization of brain/blood metabolites associated with AD-related traits (FDR≤10%). Each column represents a phenotype, with yellow color bar indicating cross-sectional AD-related traits at time of death and grey color bar indicating longitudinal cognitive phenotypes. Each row represents a metabolite, with orange color bar indicating brain metabolites; blue color bar indicating serum metabolites; and purple color indicating metabolites that exist in both brain and blood. The total number of significant associations are denoted in ().