The metabolic landscape of brain alterations in Alzheimer's disease

BACKGROUND: Impairment of brain glucose metabolism has been frequently described in Alzheimer's disease (AD). Moreover, the strongest predictor of the lifetime incidence of AD is the ε4 allele of APOE, a protein involved in lipid metabolism. These connections between AD and metabolism provide motivation to perform an in-depth metabolic profiling of human brain tissue for different stages of AD pathophysiology. METHOD: Brain tissue samples were obtained from the Religious Orders Study and Memory and Aging Project (ROS/MAP) at the Rush Alzheimer's Disease Center. Furthermore, ROS/MAP collects extensive phenotyping of the participants' cognitive trajectories as well as postmortem pathology. Metabolic profiling was performed on Metabolon's untargeted platform, yielding 1,055 quantified metabolites. Generalized linear models with appropriate linkage functions for continuous or categorical AD-related phenotypes were used to discover the association of metabolic profiles with AD-related phenotypes, such as amount of amyloid and tangles in brain, global burden of pathology, NIA-Reagan score, diagnosis (derived from Braak and CERAD scores), clinical diagnosis at the time of death, global cognition assessed during the visit before death, estimated decline of global cognition over lifetime. The models included confounder correction for age, gender, body mass index, years of education, post mortem interval, number of APOEε4 alleles, and medications. RESULT: We found 263 metabolites to be significantly associated (adjusted p-value <0.05) with one of the AD phenotypes. 137 of these metabolites were significantly associated with three or more phenotypes. Of these, nine could be replicated using an independent autopsy cohort from Mayo Clinic, five could also be replicated using a published study based on Baltimore Longitudinal Study of Aging cohort. The associated metabolites are involved in various metabolic processes known to be involved in AD pathogenesis, such as amino acid metabolism, urea cycle, and polyamine metabolism. In addition, we have identified several novel associations that could uncover the interdependence of different AD-associated metabolic processes. CONCLUSION: We have generated a comprehensive landscape of AD-associated metabolites and associated processes. These will be instrumental to fill the gap in our understanding of the metabolic components of AD pathophysiology.

Duke Scholars

Author Matthias Arnold Psychiatry & Behavioral Sciences, Behavioral Medicine & Neur ...

Author Rima Fathi Kaddurah-Daouk Psychiatry & Behavioral Sciences, Behavioral Medicine & Neur ...