Metabolomic Alterations in the Blood and Brain in Association with Alzheimer’s Disease: Evidence from in vivo to Clinical Studies

Abstract

Alzheimer’s disease (AD) has become a major health problem among the elderly population. Some evidence suggests that metabolic disturbance possibly plays a role in the pathophysiology of AD. Currently, the study of metabolomics has been used to explore changes in multiple metabolites in several diseases, including AD. Thus, the metabolomics research in AD might provide some information regarding metabolic dysregulations, and their possible associated pathophysiology. This review summarizes the information discovered regarding the metabolites in the brain and the blood from the metabolomics research of AD from both animal and clinical studies. Additionally, the correlation between the changes in metabolites and outcomes, such as pathological findings in the brain and cognitive impairment are discussed. We also deliberate on the findings of cohort studies, demonstrating the alterations in metabolites before changes of cognitive function. All of these findings can be used to inform the potential identity of specific metabolites as possible biomarkers for AD.

Keywords

Aging Alzheimer’s disease dementia metabolomics

INTRODUCTION

As the world population approaches an aging society, dementia is becoming a major health problem. More than forty-seven million people have been affected by dementia worldwide and this number is expected to increase [1]. Among the number of dementia cases, Alzheimer’s disease (AD) is associated with two-thirds of dementia cases [2]. AD is characterized by the deposition of extracellular amyloid-β protein (Aβ) as neuritic plaques and the intracellular accumulation of phosphorylated tau as neurofibrillary tangles [3]. Neuronal loss and reactive microgliosis are additional pathologic changes in the AD brain [3]. The exact pathophysiology of AD is still elusive, but several risk factors, in addition to aging, also enhance the risk of AD, including genetic risk factors such as apolipoprotein E4 (APOE4), as well as type 2 diabetes, obesity, dyslipidemia, andmetabolic syndrome [2]. These AD risk factors also involve metabolic dysfunction, which causes alterations of various metabolites in the body. Therefore, studies in metabolic changes could reveal some underlying mechanisms related to metabolic dysregulation and could discover some novel biomarkers or early therapeutic targets for dementia AD in an aging population.

To study the metabolic changes in AD, many metabolomic studies in these conditions have been conducted. The studies of the metabolome or metabolomics have revealed the simultaneous disturbance of multiple metabolites which has led the investigators to identify and observe the ongoing occurrence of multiple processes during the progression of the disease. The short details of metabolomic studies have been reviewed elsewhere [4]. Briefly, the metabolites are extracted from the sample such as blood or organ tissues, and then undergo analyses via mass spectrometry or nuclear magnetic resonance spectroscopy (NMR). The separation and identification of the metabolites are based on the mass per charge ratio (m/z ratio). Due to the large amount of data accrued from metabolomic studies, many forms of statistical analyses have been utilized to understand those data. However, due to the diversity of metabolites, the findings have been dissimilar among the studies. The different models used in the animal studies and groups of diagnosed patients have also contributed to the inconsistency of the data. Thus, this present review aims to collect and summarize the findings from metabolomic studies in dementia and AD from both in vivo and clinical studies. The controversial findings will also be discussed.

The original articles published on the PubMed database before September 2020, selected for relevance, are included in this review. The MeSH terms ‘metabolomic’, ‘dementia’, and ‘aging’ were used to search for articles. Fifty-five original articles were screened and 22 of these articles were excluded as they were not related to metabolomics research or dementia and aging. The remaining 33 references which were selected on the basis of the relevance to the topic of this review were included. The article selection method is displayed in Fig. 1.

Fig. 1

Article selection method.

ALTERATIONS IN METABOLITES FROM AD MODELS IN ANIMAL STUDIES

Phospholipids

Phospholipids are one of the major constituents of the cell membrane. The myelination of neurons in the central nervous system also plays an important role in signal transduction and this myelination also depends on normal lipid metabolism, including that of phospholipids [5]. Therefore, the alterations in phospholipids and the association with the development of AD have been a focus in several studies (Table 1) [6 –11]. The increases in the brain sphingomyelins (SM), one of the phospholipids, have been shown in genetically modified AD mice, including AβPP/PS1 mice, and PBL4 mice at the age of 8 months [6, 7]. The PBL4 mice are created by the hbace1 knock-in process which makes these mice able to produce human Aβ protein precursor (AβPP) while AβPP/PS1 mice are double transgenic mice which produce both humanized AβPP and PSEN1. The Aβ plaques develop in the brain of AβPP/PS1 mice at six months of age whereas they do not in PBL4 mice, but rather produce Aβ oligomers instead. The models of mice with elevated AD risk factors, including a model of chronologically aged female mice and a model of human APOE4 knock-in mice, showed the increases in brain ceramide and SMs, respectively [8, 9]. A human APOE4 knock-in mice model is an animal model containing a humanized APOE4 allele as an AD risk factor whilst preserving locomotor activities and working memory. The other studies raised inconsistent findings regarding several glycerophospholipids, mainly phosphatidylcholine (PC). In one study, the PCs were found to be increased at 8 months of age in AβPP/PS1 mice [6]. In another study, young PBL4 mice had reduced the brain PC levels, while some subtypes of PC were elevated in old PBL4 mice [7]. These different metabolomic profiles may be associated with the different types of Aβ generated in each model. PBL4 mice exhibited the accumulation of Aβ^*56 and Aβ hexamers in the brain and with sparse development of Aβ plaque [12], whereas AβPP/PS1 mice produced Aβ_1–42 which led to Aβ plaque formation [13]. The levels of LysoPC were also found to be elevated in the brains of aged human APOE4 knock-in female mice [9]. These data suggested that the perturbations of phospholipid metabolisms in the brain might be associated with AD.

Table 1

Changes in brain and blood metabolites in animal models with dementia

	Cognitive function	Methods	Major metabolic disturbance		Interpretations	References
			Lipids	Amino acids, polyamines, and other metabolites
Brain
Female AβPP/PS1 mice versus age-matched WT mice•6 months•8 months•10 months•12 months•18 months	NA	Targested (UPLC-MS for amino acids and biogenic amine and FIA-MS for others)	↑ PCs (8 months)↑ SMs (8 months)	Polyamines↑ arginine (6 months)↑ putrescine (6 months)↑ spermine (8 months)↑ spermidine (8 months)Amino acid↑ Phenylalanine (6 months)↑ Tryptophan (6,18 months)↓ Tyrosine (12 months)	Increases in metabolites of lipid pathway including PCs and SMs in 8-month-old mice, the polyamine pathway including arginine and putrescine in 6-month-old mice and spermine and spermidine in 8-month-old mice were observed in the brain of AβPP/PS1 mice. The metabolites in the amino acid pathway were also perturbed at 6, 12, and 18 months.	[6]
Homozygous PBL4 female mice versus WT mice•Young (4–6 months)•Old (8 months)	NA	Targeted (UPLC-MS for amino acids and biogenic amine, and FIA-MS/MS for others)	↓ PCs↑ SMs (old PBL4 mice)	Polyamine↓ putrescine	Decreased PCs in lipid pathway and putrescine in the polyamine pathway, but increased SMs in lipid pathway were found in the brain of aged PBL4 mice.	[7]
Male SAMP8 mice compared between•3 months•10 months	↑ latency time to escape the box in reversal phase of Barnes maze↓ recognition index in the novel object recognition test↑ time in open arms in the elevated plus-maze	Targeted (LC-MS/MS, and GC-MS/MS)	NA	Glycolysis and TCA cycle↑ 3-phosphoglycerate↑ 2-phosphoglycerate↑ Phosphoenolpyruvate↓ Pyruvate↑ acetyl-CoA↑ citrate↑ isocitrate↓ glutamate↓ alpha-ketoglutarate↓ NAAG↑ cAMP↓ 2-hydroxybutyrate↑ Ergothioneine	Alterations in metabolites of glycolysis and TCA cycle, as indicated by decreases in glutamate and alpha-ketoglutarate and increases in cAMP were shown in the brain of SAMP8 mice with cognitive impairment.	[19, 20]
Aged human APOE-targeted replacement mice compared between (14-15 months)•APOE3/APOE3•APOE3/APOE4•APOE4/APOE4	NA	Untargeted and targeted (LC-MS)	↓ fatty acids↑ DHA	TCA cycle↑ vitamin derivatives↑ C₆H₁₂O₆ subunit-containing - oligosaccharides↑ TCA cycle metabolites↑ mitochondrial ammonia metabolites↑ mitochondrial ammonia metabolites	Decreases in metabolites of the lipid pathway including fatty acids (except DHA) and increases in metabolites of the TCA cycle were found in the brain of APOE4 mice.	[18]
Female mice compared between•Reproductive competent (3–6 months)•Reproductive irregular (9 months)•Reproductive incompetent (12 months)•Aged (15–18 months)	Reproductive incompetent and aged mice↓ myelin basic protein level↑ % of compromised axon in the Schaffer collateral pathway and Corpus Callosum in electron microscopic analysis	Targeted (UPLC-MS for ceramide, LC-MS for fatty acids, and GC-MS for TCA metabolites)	Reproductive incompetent↑ ceramideAged↑ ceramide↑ fatty acid↑ DHA	Reproductive irregular↑ ketoneReproductive incompetent↑ ketoneAged↑ ketone↑ succinate↑ citrate	Increases in metabolites of lipid pathway, including ceramide and fatty acids, and increases in metabolites of TCA cycle and ketone body were found in the brain of reproductive incompetent and aged female mice. Those mice had myelin degeneration which represents brain aging.	[8]
Aged human APOE knock-in mice compared to•hAPOE3 male mice•hAPOE4 male mice•hAPOE3 female mice•hAPOE4 female mice	↑ Aβ 42/40 ratio in hAPOE4	Targeted (UPLC-MS/MS)	hAPOE4↑ LysoPCsMale↑ LysoPCs	hAPOE4 female↑ amino acids↑ carnitinehAPOE4 male↑ amino acids↑ carnitineMale↑ amino acids	APOE4 genotypes showed higher amino acid, carnitine, and LysoPCs metabolites in the brain. The amino acids and LysoPCs were found to be higher in males than females.	[9]
Blood
Female AβPP/PS1 mice versus age-matched WT mice•6 months•8 months•10 months•12 months•18 months	NA	Targeted (UPLC-MS for amino acids and biogenic amine and FIA-MS for others)	↓ PCs (8 months)↓ acylcarnitine C4:1, C6:1 (8,10,12,18 months)	Amino acid and polyamine↑ methionine sulfoxide (6, 10 months)↑ arginine (6 months)↓ EAA: NEAA (6, 12, 18 months)↑ serotonin (8 months)↑ tyrosine (8 months)↑ group of the amino acid (Proline, Serine, Asparagine, Phenylalanine, Histidine, Tyrosine, Taurine) (10 months)↑ polyamine (putrescine, spermine, spermidine) (10 months)↑ methionine (10 months)↑ isoleucine (12 months)↓ glutamine, citrulline, alpha-aminoadipic acid, valine, methionine, threonine (12 months)	Decreased metabolites of lipid pathway including PCs and acylcarnitines in the blood of 6- to 18-month-old AβPP/PS1 mice were observed. Increases in the polyamines in the blood after the increased polyamines in the brain suggested that the pathology might initially occur in the brain before a whole body system. The amino acid pathway was altered throughout all age groups.	[6]
Male 12-week-old Kunming mice•Normal•Control injection•Aβ_1-42 injection into hippocampus	↑ time to reach platforms↓ time in the target quadrant↓ number of platform crossing	Targeted (UPLC-MS/MS)	↓ dihydrosphingosine↓ LysoPCs	↑ phenylalanine↓ tryptophan	Decreased dihydrosphingosine LysoPCs, tryptophan, as well as increased phenylalanine were found in demented mice induced by Aβ_1-42 injection.	[10]
Male SAMP8 mice compared between•3 months•10 months	↑ Latency time to escape the box in reversal phase of Barnes maze↓ Recognition index in the novel object recognition test↑ Time in open arms in the elevated plus-maze	Targeted (LC-MS/MS, and GC-MS/MS)	↑ acylcarnitines↑ PUFAs	↓ γ-glutamyl amino acids↓ BCAAs↓ dipeptides↓ 4-hydroxyphenylacetate↓ Indoleacetate↓ Homocitrulline↑ Caprylate↓ 4-sulfooxymethylbenzoate	The increases in acylcarnitines and PUFAs and the decreases in metabolites in the amino acid pathway were found in aged SAMP8 mice.	[19, 20]
16-month-old mice compared between•hAPOE3 male•hAPOE4 male•hAPOE3 female•hAPOE4 female	↑ Aβ 42/40 ratio in hAPOE4	Targeted (UPLC-MS/MS)	hAPOE4 female↑↑ acylcarnitine↑ SMhAPOE4 male↑ PCs↑ LysoPCsFemale↑ acylcarnitineMale↑ PCs↑ LysoPCs	hAPOE4 male↑↑ glucogenic amino acidMale↑ glucogenic amino acid	The glucogenic amino acid and some lipid metabolites such as PCs and LysoPCs were higher in male mice, while the acylcarnitine was higher in female mice. Interestingly, the APOE4 genotype further increased the glucogenic amino acid in male mice and acylcarnitine in female mice.	[9]
ApoE-targeted replacement miceAge group•3 months•12 months•24 monthsAPOE genotype•APOE2•APOE3•APOE4Sex•Male•Female	NA	Targeted (UPLC-MS/MS and FIA-MS/MS)	APOE2↑ phospholipid↑ sphingolipid12 and 24 months↑ acylcarnitineFemale↓ PCs↓ LysoPCs12 and 24 months (APOE3,4)↓ LysoPCs	12 and 24 months↓ amino acids↓ biogenic amines	Increased acylcarnitine, as well as decreased amino acids and biogenic amines, were found in aged mice, but increased phospholipid and sphingolipid levels were observed in APOE2 genotype mice.	[11]

Aβ, amyloid β; APOE, apolipoprotein E; BCAAs, branched-chain amino acids; cAMP, cyclic adenosine monophosphate; CoA, coenzyme A; DHA, docosahexaenoic acid; EAA, essential amino acids; FIA-MS/MS, flow-injection - tandem mass spectrometry; GC-MS, gas chromatography-mass spectrometry; GC-MS/MS, gas chromatography-tandem mass spectrometry; hAPOE, human APOE; LC-MS, liquid chromatography-mass spectrometry; LC-MS/MS, liquid chromatography-tandem mass spectrometry; LysoPCs, Lysophosphatidylcholines; NA, not available; NAAG, N-acetyl-aspartyl-glutamate; NEAA, non-essential amino acids; PCs, phosphatidylcholine; PUFAs, polyunsaturated fatty acids; SAMP8, senescence-accelerated prone 8; SMs, sphingomyelin; TCA, tricarboxylic acid; UPLC-MS, ultraperformance liquid chromatography-mass spectrometer; UPLC-MS/MS, ultraperformance liquid chromatography-tandem mass spectrometer; WT, wild type.

Changes in phospholipid levels were also observed in plasma; however, these were not associated with the changes in the brain. In AβPP/PS1 mice, the level of PCs in plasma was decreased at the age of 8 months [6]. Additionally, the reductions of LysoPC in plasma were observed in the mice with hippocampal Aβ_1-42 injections [10]. The acute Aβ_1-42 injections into the hippocampus result in impaired learning and memory processes, and neurodegeneration which resembles AD. This model could demonstrate the direct effects of Aβ_1-42 on the brain and cognitive function of the animal. Conversely, the PCs and LysoPCs were relatively higher in the male mice when compared to the female mice despite being of the same APOE genotype [9]. The lower PCs and LysoPCs in female mice might be associated with AD because the incidence of AD was slightly higher in females than in males [14]. Additionally, the mice with APOE2 replacement, which showed beneficial effects on cognitive function, had increases in the levels of phospholipid in the plasma [11]. These data suggested that the elevations of phospholipid in the plasma of male and APOE2 replacement mice might be inversely associated with AD. In the case of sphingolipid in the plasma, dihydrosphingosine was decreased in mice after hippocampal Aβ_1-42 injections [10]. The associations of sphingolipid changes in plasma and AD need to be investigated further.

In summary, the phospholipid changes in the brain and plasma in animal models were associated with AD. The PCs tended to be elevated in the brain from AD models, while the circulating PCs may confer beneficial effects. Increased sphingolipid levels in the brain were also associated with AD. The dysregulation of phospholipids may be associated with Aβ formation and aggregation, impairment of synaptic activity, and neuronal cell death, resulting in the development of AD [15].

Fatty acids

The brain is a metabolically active organ, that utilizes glucose as a major source of energy [16]. Although fatty acid can be an alternative energy source for brain metabolism, there are still some disadvantages [17]. Firstly, to achieve the same level of energy production, more oxygen is required for fatty acid oxidation than glucose metabolism which may lead to tissue hypoxia of the brain [17]. Secondly, the metabolism of fatty acid through the β-oxidation process and the electron transport chain of mitochondria produced a large number of oxidative stress [17]. Finally, the rate of ATP generation from the fatty acids might be too slow and could not meet the usual metabolic demand of the brain since it depends on long and complex metabolism pathways [17]. Therefore, the disturbances in fatty acid metabolism, as indicated by altered levels of fatty acid and acylcarnitine, may be associated with AD (Table 1) [6 , 18–20]. The brain of aged human APOE4 knock-in mice showed the decreased fatty acid levels indicating utilization of fatty acids as a fuel source by neurons during the neuronal hyperactivity stage, which was demonstrated by the upregulation of tricarboxylic acid (TCA) cycle metabolites including citrate, isocitrate, and malate [18]. However, in the brain of aged female mice without any AD pathology or APOE risk factors, the fatty acids were increased as a consequence of sphingomyelinase activity [8]. The alterations in metabolites in the TCA cycle were observed in SAMP8 mice [20]. SAMP8 mice are naturally occurring mice which exhibit phenotypes of accelerated aging and neuropathology such as microgliosis, neurodegeneration, and hippocampal Aβ without plaque formation. Higher levels of polyunsaturated fatty acids (PUFAs) were also found in the blood of SAMP8 mice [20]. In addition to fatty acid, the changes observed in the levels of acylcarnitine might reflect the abnormal lipid metabolism. In the female AβPP/PS1 mice, the acylcarnitine levels were altered in the brain indicating the dysregulation of lipid metabolism in the brain in cases of AD [6]. Furthermore, aged-SAMP8 mice and human APOE4 knock-in female mice showed increased acylcarnitine levels in their blood [9 , 20]. Acylcarnitines were also found to be higher in female or aging mice in these studies [9, 11]. However, the carnitine level was increased in both male and female human APOE4 knock-in mice [9]. These findings suggested that the disturbances in the lipid metabolism in relation to increased acylcarnitine levels were associated with risk factors for AD. However, in the female AβPP/PS1 mice, the acylcarnitine C4:1 and C6:1 levels were reduced in the blood during the period of 8–18 months of age when compared to their wild-type counterpart at each age [6]. These different findings might be due to differing pathologies of AD in each study.

In conclusion, changes in fatty acid metabolism were found the brain and the blood of AD models in animal studies. The elevation in fatty acid levels and acylcarnitine levels potentially reflect the increase in the metabolic demand associated with AD or the risk factors of AD.

Amino acids

Amino acids play an important role in the function of the brain as the precursors of several neurotransmitters [21]. Some metabolomic research on AD has focused on the changes in these metabolites (Table 1) [6 , 20]. Overall changes in amino acids were observed in the brain of human APOE4 knock-in mice. Male APOE4 mice had higher levels of glucogenic amino acid in the brain than the female mice which had a higher level of LysoPCs. This information suggested that the female mice, which are associated with a higher risk of dementia and AD, tended to exhibit a lipid metabolic profile [9]. Furthermore, the decreases in amino acids and biogenic amines in the blood were associated with the age in one animal metabolomic study [11]. These findings suggested that, overall, the increased levels of several amino acids were inversely associated with dementia and AD. The metabolomic study in AβPP/PS1 mice showed transient changes in several amino acids in the different ages both in the blood and brain. These indicated the different precipitating factors of AD including impairment of amino acid management by the liver, disturbance of neurotransmitter function, or increased oxidative stress [6]. Additionally, changes in amino acid levels in the plasma were observed in several studies, suggesting changes in both metabolisms and the oxidative state in the AD model [10 , 20].

Polyamine is a biogenic amine derived from the amino acid arginine and is involved in diverse physiological functions [22]. The polyamines were increased in the model of AβPP/PS1 mice at the ageof 6 and 8 months, while the changes in the bloodwere only observed at around 10 months in the same model [6]. An increase in polyamine might promote the higher activity of the N-methyl-D-aspartate (NMDA) receptor, leading to neuronal excitotoxicity. However, in the PBL4 mice, the level of putrescine, one of the polyamines, was decreased [7]. The differences in the changes between these two studies might be explained by the differences in Aβ formation in different AD models.

Conclusively, the alterations in amino acids and biogenic amines such as polyamine could reflect several pathophysiologies of AD. The dynamic changes of the disease might lead to the fluctuation of the levels of these amino acids and polyamine metabolites which made the findings inconsistent among the studies.

METABOLIC CHANGES IN ANIMAL MODELS WITH AD FOLLOWING NEUROPROTECTIVE INTERVENTIONS

Several neuroprotective compounds were used in these models of dementia and AD to improve cognitive function and brain pathology. Thoseinterventions altered the level of the metabolites both in the brain and the blood (Table 2) [10 , 19,20]. J147 is the neuroprotective compound derived from curcumin and provides beneficial effects on cognitive function [23]. J147 treatment altered themetabolites in SAMP8 mice as shown by decreases in acylcarnitine and PUFAs, and increases in aminoacids in the blood including the gamma-glutamyl amino acids, branched-chain amino acid, and dipeptides, which may be associated with the improvement of the metabolism and oxidative status [20]. Additionally, the upregulations of glutamate and alpha-ketoglutarate and a decrease in cAMP in the brain were also observed after the J147 treatment indicating metabolic improvement [20]. Fisetin, a bioactive flavonoid, also provides neuroprotective effects via multiple mechanisms [24]. Treatment with fisetin improved cognitive function and changed the metabolites in the brain and blood of old SAMP8 mice to be similar to metabolites those in the young SAMP8 mice. The differences in changed metabolites between Fisetin and J147 might be due to the different mechanisms of each compound [19, 20]. GRb1 and GRg1, the major active ingredients of ginseng were administrated in 12-week-old mice with hippocampal Aβ_1-42 injection. These treatments showed neuroprotective effects and improved the metabolites in the plasma [10]. The high dosage of GRb1 and GRg1 decreased the level of phenylalanine but increased the levels of tryptophan and LysoPCs in plasma [10]. However high dosage GRg1, but not GRb1 increased the dihydrosphingosine level in plasma [10]. This inconsistency might be due to the different chemical structures of these two compounds.

Table 2

Change in brain and blood metabolites in animal models with AD following neuroprotective interventions

	Treatment (dose/time)	Methods	Major findings		Interpretation	References
			Parameter changes	Metabolites changes
Male 3-month-old SAMP8 mice•Control•treated with a control diet for 7 months•treated with J147 for 7 months	J147200 ppm (∼10 mg/kg/day, PO)	Targeted (LC-MS/MS, and GC-MS/MS)	↓ Aβ 1-40, tau↑ locomotor activity↑ elevated plus-maze test↑ object recognition test↑ Barnes maze test	Brain↑ glutamate↓ cAMP↑ alpha-ketoglutaratePlasma↓ acylcarnitine↓ PUFAs↑ γ-glutamyl amino acids↑ BCAAs↑ dipeptides	J147 improved cognitive function in male SAMP8 mice and was also associated with the increases in glutamate and alpha-ketoglutarate in the brain and peptide and amino acid in plasma and the decreases in cAMP in the brain and acylcarnitine and fatty acid in plasma.	[20]
Male 3-month-old SAMP8 mice•control•treated with a control diet for 7 months•treated with Fiestin for 7 months	Fisetin500 ppm (∼25 mg/kg/day, PO)	Targeted (LC-MS/MS, and GC-MS/MS)	↑ locomotor activity↑ object recognition test↑ Barnes maze test	Brain↑ 2-hydroxybutyrate↑ Trans-4-hydroxyproline↑ ribitol↓ ergothioneinePlasma↑ 4-hydroxyphenylacetate↑ indoleacetate↑ homocitrulline↓ caprylate↑ 4-sulfooxymethylbenzoate	Fisetin improved cognitive function in male SAMP8 mice which is partially associated with alteration of metabolites in the brain and plasma.	[19]
Male 12-week-old Khun ming mice injected with Aβ_1-42 into the hippocampus•Control•GRb1 low dose for 1 month•GRb1 moderate dose for 1 month•GRb1 high dose for 1 month	GRb1Low dose (7.5 mg/kg/day, IP)Moderate dose (15 mg/kg/day, IP)High dose (30 mg/kg/day, IP)	Targeted (UPLC-MS/MS)	↓ Aβ_1-42↑ Morris water maze test	Plasma↓ phenylalanine↑ tryptophan↑ LysoPCs	GRb1 improved cognitive function and brain Aβ deposition and was associated with increases in tryptophan and LysoPCs and the decrease in phenylalanine in plasma.	[10]
Male 12-week-old Khun ming mice injected with Aβ_1-42 into the hippocampus•Control•GRg1 low dose for 1 month•GRg1 moderate dose for 1 month•GRg1 high dose for 1 month	GRg1Low dose (7.5 mg/kg/day, IP)Moderate dose (15 mg/kg/day, IP)High dose (30 mg/kg/day, IP)	Targeted (UPLC-MS/MS)	↓ Aβ_1-42↑ Morris water maze test	Plasma↓phenylalanine↑tryptophan↑dihydrosphingosine↑LysoPCs	GRg1 improved cognitive function and brain Aβ deposition and was associated with the increase in tryptophan, dihydrosphingosine, and LysoPCs and the decrease in phenylalanine in plasma.	[10]

Aβ, amyloid β; BCAAs, branched-chain amino acids; cAMP, cyclic adenosine monophosphate; GC-MS/MS, gas chromatography-tandem mass spectrometry; IP, intraperitoneal; LC-MS/MS, liquid chromatography-tandem mass spectrometry; LysoPCs, Lysophosphatidylcholines; PO, peroral; PUFAs, polyunsaturated fatty acids; UPLC-MS/MS, ultraperformance liquid chromatography-tandem mass spectrometer; SAMP8, senescence-accelerated Mice P8.

In summary, the interventions that improve cognitive function and brain pathology also change the level of several blood and brain metabolites of AD models. The inconsistent changes in metabolites between studies might be due to the different models of dementia and the different mechanisms of the interventions. A study using a single model with an intervention with a specific target should be further investigated.

CHANGES IN METABOLITES ASSOCIATED WITH AD: EVIDENCE FROM CLINICAL STUDIES

Some of the clinical studies used a metabolomic approach to compare the profiles of brain metabolites between normal subjects and AD patients (as shown in Table 3) [25 –30]. The authors demonstrated that the changes in sphingolipids, glycerophospholipids, and fatty acids levels in the brain were associated with AD clinical status. Those changes were associated with the severity of AD, as indicated by changes in sequence from AD > asymptomatic AD > normal [25, 26]. A reduction in brain glucose metabolism, as indicated by decreases in the ratios of brain metabolites in glycolysis, was also observed in progressive AD [27]. Additionally, the disturbances of several amino acids and biogenic amines in the AD brain indicated that the transmethylated process was dysregulated in cases of AD [28]. The changes in the levels of sphingolipids, phospholipids, fatty acids, glucose, amino acids, and polyamines in these studies were associated with brain AD pathologies, measurable using the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) and Braak scores, which are used to represent the neuritic plaque burden and level of neurofibrillary tangle in the brain, respectively. Increases in SDMA and threonine levels were also associated with brain-related AD phenotypes [31]. In addition to the brain, an untargeted-metabolomic study of cerebrospinal fluid (CSF) also detected changes in the levels of metabolites in AD patients, when compared with non-AD controls [29]. Another study into CSF in dementia patients demonstrated a reduction in metabolites related to polyamine and tryptophan-kynurenine [30]. All of these findings from metabolomics research in the human brain revealed the dysregulation of the metabolites which may be related to cell membrane integrity, alteration in the source of energy, and disturbance in neurotransmitter metabolism. Furthermore, the impairment of the blood-brain barrier which occurs during the progression of AD might lead to the differing composition of the metabolites in the brain and CSF. The study of ratios of the metabolites between these two areas might provide useful information regarding blood-brain barrier function.

Table 3

Changes in brain, CSF, and blood metabolites in patients with cognitive decline and associated brain pathologies

	Methods	Major metabolic disturbance		Associated AD Pathologies	Interpretation	References
		Lipid	Amino acid, polyamine, and other metabolites
Brain
Normal controls (n = 14, mean age 80±11)Asymptomatic AD (n = 15, mean age 85±9)AD (n = 15, mean age 78±10)	Targeted (FIA-MS/MS for lipids)	AD > asymptomatic AD > Normal controls↑ SMs (SM C16:0, SM C16:1, SM (OH) C14:1)↓ glycerophospholipid (PC ae C36:0, PC ae C40:1, PC aa C40:4)	NA	The severity of neuritic plaque burden (CERAD score) and Neurofibrillary pathology (Braak score)	SMs were increased and the glycerophospholipid was decreased in the brain of AD patients and these were also associated with AD pathologies.	[25]
Normal controls (n = 14, mean age 83±11)Asymptomatic AD (n = 15, mean age 89±8)AD (n = 14, mean age 88±9)	Targeted (FIA-MS/MS for glucose and LC-MS/MS for amino acids)	NA	AD > asymptomatic AD > Normal controls↑ glucose↓ serine + glycine/glucose↓ serine + glycine + alanine/glucose	The severity of neuritic plaque burden (CERAD score) and Neurofibrillary pathology (Braak score)	An increase in glucose accumulation along with a reduction in glycolytic flux was observed in AD patients and associated with AD pathologies.	[27]
Normal controls (n = 14, mean age 83±11)Asymptomatic AD (n = 15, mean age 89±8)AD (n = 14, mean age 88±9)	Untargeted (GC-MS, LC-MS)	AD > asymptomatic AD > Normal controls↓ linoleic acid↓ linolenic acid↓ EPA↓ oleic acid↓ arachidonic acid↑ DHA	NA	The severity of neuritic plaque burden (CERAD score) and Neurofibrillary pathology (Braak score)	The fatty acids were found to be decreased in the brain of AD patients (except DHA) and these were also associated with AD pathologies.	[26]
Normal controls (n = 13, mean age 82±12)Asymptomatic AD (n = 13, mean age 88±8)AD (n = 17, mean age 87±9)	Targeted (CE-MS)	NA	Polyamine↑ spermidineMethionine cycle↓ choline↑ SAMTranssulfuration and glutathione synthesis↑ cysteine↑ GSHUrea cycle↓ NAGGlutamate aspartate metabolism↓ NAANeurotransmitter metabolism↓ GABA	The severity of neuritic plaque burden (CERAD score) and Neurofibrillary pathology (Braak score)	Dysregulation in transmethylation and polyamine pathway as indicated by changes in metabolites in the methionine cycle, transsulfuration and glutathione synthesis, polyamine metabolism, urea cycle, glutamate-aspartate metabolism, and neurotransmitter metabolism were found in AD patients and these were also associated with AD pathology.	[28]
Normal controls (n = 83, mean age = 90±5)AD (n = 28, mean age = 81±6)	Targeted (LC-MS/MS for amino acids and biogenic amines)	↑ SDMA↑ threonine	Brain-related AD phenotypes.(autopsy-confirmed AD diagnosis, CERAD score, global AD pathology burden, Aβ, neuritic plaque)	Increases in SDMA and threonine were associated with AD pathologies in the brain.	[31]
CSF
Normal controls (n = 38, mean age 70±10)AD (n = 40, mean age 69±10)	Untargeted (GC-MS)	NA	↑ 15-65.533 (m/z)↑ 8-93.65 (m/z)	NA	Changes in metabolites in CSF were observed in the patients with AD pathology.	[29]
Normal controls (n = 10, mean age 87±3)AD (n = 10, mean age 88±4)	Targeted (UPLC-MS/MS)	NA	Polyamine↓ ornithine↓ diacetyl-spermidine↓ diacetyl-spermineTryptophan-kynurenine↓ tryptophan↓ kynurenine↓ phenylalanine↓ anthranilate↓ 3-hydroxy kynurenineOther metabolites↑ methionine sulfoxide	NA	Metabolites in the polyamine pathway and tryptophan-kynurenine pathway were found to be decreased in the CSF of AD patients.	[30]
Blood
Normal controls (n = 216, mean age 76±5)MCI (n = 366, mean age 75±7)AD (n = 185, mean age 75±7)	Targeted (FIA-MS/MS for lipids)	↑ SM C18:1↑ SM C16:0↓ PC aa C40:6	NA	Brain pathology from MRI (SPARE-AD index)	Increases in SMs and a decrease in PC were associated with brain pathology.	[25]
Normal controls (n = 216, mean age 76±5)MCI (n = 366, mean age 75±7)AD (n = 185, mean age 75±7)	Targeted (FIA-MS/MS for acylcarnitines and lipids and LC-MS/MS for amino acids and biogenic amines)	↑ SM C16:0↑ SM (OH) C14:1↑ SM C16:0↑ SM (OH) C14:1↓ SM C16:0↓ SM (OH) C14:1↓ C3 acylcarnitine	↓ Serotonin	CSF p-tauCSF t-tauCSF Aβ_1-42	Increases in SMs associated with tau protein in CSF and decreases in SMs, C3 acylcarnitine, and serotonin were associated with Aβ _1-42 in CSF.	[25]
Low NAB (n = 48, median age (IQR) = 66 (9))High NAB (n = 30, median age (IQR) = 63 (13))	Targeted (UPLC-MS/MS)	↓ PE 39:7↑ PC aa 36:6	↑ Anandamide isotope↑ Anandamide	Prediction of amyloid positivity	Increases in PC and anandamide and a decrease in PE 39:7 were associated with brain amyloid pathology.	[44]
Normal controls (n = 17, age 68)MCI (n = 10, age = 72)AD (n = 7, age = 72)	Targeted (UPLC-MS/MS)	↓ PC ae 42:4↓ PC ae 44:4↓ SM (OH) C14:1	NA	Amyloid positivity on PET scan	Decreases in PCs and SMs were associated with brain amyloid pathology.	[43]
Normal controls (n = 199, age 75 (72-78))MCI (n = 358, age 75 (70–80))AD (n = 175, age 76 (71–80))	Targeted (UPLC-MS/MS)	↑ PC ae C40:3,↑ PC ae C42:4↑ PC ae C44:4↑ SM (OH) C14:1↑ SM C16:0↑ SM C20:2↑ C18↑ PC ae C36:2↑ SM C16:0↑ SM C20:2↑ PC ae C36:2↑ PC ae C40:3↑ PC ae C42:4↑ PC ae C44:4↑ SM (OH) C14:1↑ SM C16:0	↓ valine	Ventricular volume changes up to 5 yearsTau/Aβ_1-42 ratioCSF Aβ_1–42	Decreases in PCs and SMs and an increase in valine were associated with lower ventricular volume.Increases in SPHs, PCs, and acylcarnitine were associated with tau protein in CSF, and increases in PCs and SPHs were associated with Aβ _1-42 in CSF.	[43]
Normal controls (n = 370, mean age 75±6)Early MCI (n = 284, mean age 71±8)Late MCI (n = 505, mean age 74±8)AD (n = 305, mean age 75±8)	Targeted (LC-MS/MS, UPLC-MS/MS)	–	AD vs MCI or Normal controls Primary bile acid↓ CASecondary bile acid↑ DCA↑ GDCA↑ TDCA↑ GLCAAD vs Normal controls Secondary/primary bile acid ratio↑ DCA:CA↑ GDCA:CA↑ TDCA:CA↑ GLCA:CDCA	NA	Increases in secondary bile acids and the secondary/primary bile acid ratio were found in AD patients, indicating that changes in gut microbiota were associated with the development of AD.	[40]
MCI (n = 40, median age 68 (67–72))Early AD (n = 40, median age 77 (73–81))	Targeted (MS)	↓ pimelylcarnitine↓ SM (OH) C24:1↓ SM C24:0↑ PC ae C44:3	↑ acetylornithine↑ methionine sulfoxide↓ putrescine	NA	Decreases in acylcarnitine, SMs, and putrescine and increases in PCs, acetylornithine, and methionine sulfoxide were found in early-stage AD.	[33]
Normal controls (n = 46, mean age 71±6)Stable MCI (n = 91, mean age 72±5)Progressive MCI (n = 52, mean age 71±6)AD (n = 37, mean age 75±4)	Untargeted (UPLC-MS for lipids and GC-MS for others)	AD > MCI > Normal controls ↓ ether phospholipids (Plasmalogen)↓ PCs↓ SMs↓ sterols	NA	NA	Decreases in metabolites in the lipid pathway including ether phospholipid, PCs, SMs, and sterols were found in MCI and AD patients.	[35]
Normal controls (n = 34, mean age 72±4)MCI (n = 38, mean age 72±3)AD (n = 35, mean age 73±4)	Targeted (GC-MS for 2-hydroxybutyric acid and LC-MS/MS for others)	AD > MCI or Normal controls ↓ octanoylcarnitine (C8)↓ decenoylcarnitine (C10:1)↓ decanoylcarnitine (C10)↓ dodecenoylcarnitine (C12:1)↓ lauroylcarnitine (C12)↓ tetradecenoylcarnitine (C14:1)↓ tetradecadienoylcarnitine (C14:2)↓ oleyl carnitine (C18:1)↓ acetyl carnitine (C2) AD > MCI ↓ free carnitine (C0)	AD > MCI > Normal controls ↓ 2-hydroxybutyric acid	Gray matter volume ((+) C8, C10, C12)	Medium-chain acylcarnitine, free carnitine, and ketone bodies such as 2-hydroxybutyric acid were found to be decreased in AD patients. The decreased medium-chain acylcarnitine was associated with lower gray matter volume.	[32]
Normal controls (n = 37, mean age 73±9)MCI (n = 16, mean age 72±7)AD (n = 19, mean age 78±4)	Untargeted (UPLC-MS)	NA	AD > Normal controls ↓ L-ornithine↑ N-acetylputrescine↓ 4-aminobutanal↓ GABA↑ N1, N12-diacetylspermine↑ spermine↑ creatine MCI > Normal controls ↑ L-arginine↓ L-ornithine↑ putrescine↑ N-acetylputrescine↑ 4-aminobutanal↓ GABA↓ spermidine↑ N1, N12-diacetylspermine↓ N1 or N8-acetyl-spermidine↓ methylthioadenosine AD > MCI ↑ L-arginine↓ L-ornithine↑ putrescine↓ 4-aminobutanal↓ N-acetylputrescine↑ spermine↑ spermidine↑ methylthioadenosine↑ N1 or N8-acetyl-spermidine↑ creatine	NA	Changes in metabolites in arginine and polyamine pathways were found to differ between MCI and AD patients.	[42]
Total (n = 25,872, age 40–84)Normal controls (n = 23,882)Dementia (n = 1,990)AD (n = 1,356)	Targeted (NMR)	↓ free cholesterol in small HDLs↓ DHA↓ subfraction (phospholipid, TC, cholesterol ester) of medium size HDL (dementia only)	↑ glutamine↑ ornithine	NA	Lipid metabolites such as subfractions of HDLs and DHA were decreased in both dementia and AD patients while glutamine was increased.	[38]
Normal controls (n = 21, median age (IQR) = 81(9))AD (n = 18, median age (IQR) = 78(9))	Targeted (UPLC-MS)	↑ ceramide↑ monohexosylceramide	NA	NA	Increased levels of lipid metabolites including ceramide and monohexosylceramide were observed in AD patients.	[36]
Normal controls (n = 15, mean age 67±6)MCI (n = 10, mean age 75±9)AD (n = 15, mean age 77±8)	Targeted (UPLC-MS/MS)	AD > MCI ↓ C3 acylcarnitine↓ C5 acylcarnitine	MCI > Normal controls↑ arginine↑ phenylalanine↑ creatinine↑ SDMA	NA	A decrease in acylcarnitine was found in AD patients, while increases in amino acids including arginine, phenylalanine, creatinine, and SDMA were observed in MCI patients.	[34]
Normal controls (n = 29, median age (IQR) = 65(7))MCI/AD (n = 29, median age (IQR) = 72(6))	Untargeted (UPLC-MS)	↑ Lysophosphatidylethanolamine	↑ Choline↑ Soraphen A	NA	Increases in Lysophosphatidylethanolamine, choline, and soraphen A were found in patients with MCI or AD.	[37]
Normal controls (n = 24, mean age 75±5)Dementia (n = 12, mean age 75±4)	Targeted (GC-MS)	↑ docosapentaenoic acid↑ 3,4-Dihydroxybutanoic acid	↑ uric acid	NA	Increased docosapentaenoic acid, 3,4-Dihydroxybutanoic acid, and uric acid were associated with dementia.	[39]

AD, Alzheimer’s disease; Aβ, amyloid β; CA, cholic acid; CDCA, chenodeoxycholic acid; CERAD, Consortium to Establish a Registry for Alzheimer’s Disease; CSF, cerebrospinal fluid; DCA, deoxycholic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; ESI-MS/MS, electrospray ionization - mass spectrometry; FIA-MS/MS, flow-injection - tandem mass spectrometry; GABA, gamma-Aminobutyric acid; GC-MS, gas chromatography - mass spectrometry; GC-MS/MS, gas chromatography - tandem mass spectrometry; GDCA, glycodeoxycholic acid; GLCA, glycolithocholic acid; GSH, thiol-reduced glutathione; HDL, high-density lipoprotein; HPLC-MS/MS, high-performance liquid chromatography - tandem mass spectrometry; IQR, interquartile range; LC-MS, liquid chromatography - mass spectrometry; LC-MS/MS, liquid chromatography - tandem mass spectrometry; MCI, mild cognitive impairment; NA, not available; MRI, Magnetic resonance imaging; NAA, N-acetylaspartate; NAB, neocortical Aβ burden; NAG, N-acetylglutamate; PCs, phosphatidylcholine; PE, phosphoethanolamine; PET, Positron emission tomography; SAM, S-adenosylmethionine; SDMA, symmetric dimethylarginine; SMs, sphingomyelin; SPARE-AD, Spatial Pattern of Abnormality for Recognition of Early Alzheimer’s disease; TC, total cholesterol; TDCA, taurodeoxycholic acid; UPLC-MS, ultraperformance liquid chromatography - mass spectrometer; UPLC-MS/MS, ultraperformance liquid chromatography - tandem mass spectrometer.

Most of the clinical metabolomics studies specific to AD were conducted using plasma or serum samples (Table 3) [32 –42]. However, despite the convenience of accessing the specimens, plasma or serum may not directly reflect the changes in the brain. However, information from plasma and serum contains information regarding abnormality throughout the body, which might be indirectly associated with the development of dementia. The changes in lipid metabolites from plasma and serum are the main findings from those clinical studies (Table 3) [25 , 44]. The levels of acylcarnitines and free carnitine in plasma were both reduced when the disease progressed from normal to mild cognitive impairment (MCI) or from MCI to AD [32]. In one study, acylcarnitines including C8, 10, and 12 were associated with a higher volume of gray matter in the brain [32]. Reductions in levels of acylcarnitines were also observed in another two studies, which compared the presence of metabolites in AD and MCI [33, 34]. These findings indicated that decreases in serum and plasma acylcarnitine were associated with clinical symptoms of AD. In the case of other lipids, one untargeted-metabolomic study found that the levels of ether phospholipid, PCs, SMs, and sterols were reduced in correlation with the increasing severity of AD [35]. In another study, SMs were lower and PC was higher in cases of early AD in comparison to MCI [33]. Conversely, several phospholipids were increased in AD or MCI, when compared to those of non-AD and non-MCI controls [33 , 37]. The findings within each study, even though conducted specifically in association with AD, cannot be directly compared due to the different methodologies, variety of metabolites, and clinical symptoms. In the case of fatty acid and cholesterol levels, one study found that the docosahexaenoic acid (DHA) and cholesterol level in HDL were reduced in dementia and AD patients, while another study showed that the levels of fatty acids, including docosapentaenoic acid and 3,4-dihydroxybutanoic acid, were increased in dementia patients [38, 39]. Wider age ranges and different methods were used in these two studies therefore the results are not directly comparable. For example, the NMR method was used in the first study to compare aging participants, but a gas chromatography-mass spectrometry (GC-MS) metabolomic study was used in the latter study. These differences in methodology might contribute to the different findings. Several studies also investigated the changes in blood metabolomic profiles and the associations with AD pathologies. Increases in blood SM were associated with several brain pathologies verified by MRI and the levels of tau protein and Aβ in the CFS [25, 43]. However, some SMs were inversely associated with amyloid positivity on a PET scan and CSF Aβ, indicating specific associations of these SMs to Aβ in both the brain and CSF [43]. An increase in PC in the blood was also associated with amyloid positivity [44], ventricular volume change [43], and CSF Aβ level [43]. However, some phospholipids were inversely associated with brain pathologies including PC aa C40:6 and PE 39:7[25, 44]. PC ae 42:4 and PC ae 44:4 were negatively associated with Aβ positivity on positron emission tomography (PET) scans of the brain, but were positively associated with ventricular volume changes and an increase in CSF Aβ level [43]. In addition to the phospholipids, acylcarnitine C3 was negatively associated with CSF Aβ [25]. In conclusion, the disturbances of lipid metabolites in dementia and AD patients were observed, as shown by the lowering of acylcarnitine levels and the fluctuation in PCs, SMs, and fatty acids. These findings suggested that a disruption in lipid metabolism occurred in association with AD and these changes were also related to AD pathologies. A study involving a homogenous population and using similar methods should be conducted to verify the results and identify the specific pathways responsible for these metabolic changes.

In addition to the lipids, alterations in other metabolites were also detected in several metabolomic clinical studies (Table 3) [33 , 37–42]. Increased secondary bile acid levels and a decrease in primary bile acid level in the circulation were found in AD patients when compared to MCI or control participants [40]. These findings suggested a correlation between gut dysbiosis and cognitive decline, which has been demonstrated in several metagenomic studies [41]. Changes in amino acids and biogenic amines have also been observed in the blood of AD and MCI patients. One study which compared early AD patients with MCI patients, found that the levels of acetylornithine and methionine sulfoxide were increased, while the putrescine was decreased [33]. However, another untargeted metabolomic study found that the level of putrescine was higher in MCI patients who later underwent conversion to AD when compared with normal controls or MCI patients without AD conversion [42]. The differences in the findings of these two studies could be due to the use of different metabolomic methods and the different methods of cognitive assessment used with AD patients. The higher level of cognitive function observed in the latter study may be associated with the higher putrescine level since that study demonstrated that the downstream metabolisms of putrescine differed between MCI and AD patients [42]. Another study also found that the level of arginine, the precursor of polyamines and other amino acids in MCI patients was elevated [34]. These findings indicated disturbances in amino acid and polyamine metabolism which may result in the dysregulation of neurotransmitter metabolism in AD patients. Disturbances in other metabolites including glutamine [38], choline [37], Soraphen A [37], and uric acid [39] were also observed in dementia and AD patients. Some metabolites also showed an association with AD pathologies in other studies [25 , 44]. Anandamide was positively associated with Aβ positivity in the brain [44], while valine and serotonin were negatively associated with ventricular volume change and Aβ level in CSF, respectively [25, 43].

THE ASSOCIATION BETWEEN METABOLIC ALTERATIONS AND COGNITIVE FUNCTION IN CLINICAL STUDIES

The changes in metabolites in the blood and the brain were associated with cognitive function as shown in several studies (Table 4) [25 , 45]. In the brain, the untargeted-metabolomic study found that the increases in fatty acids in the brain were associated with better cognitive function in AD patients [26]. Regarding blood metabolites, the changes in lipids, bile acids, amino acids, and other metabolites were observed. Some SMs and PCs were found to be associated with worsening cognitive function [25, 31] or increased severity of dementia [43]. However, several studies showed that increased acylcarnitines were associated with better global cognition [31, 32]. The free cholesterol level in HDL and DHA were found to be associated with the global cognitive ability [38]. In the case of other metabolites, increased secondary bile acids were associated with dementia severity [40]. Furthermore, alterations in amino acids and biogenic amines were observed. Arginine [25] and asparagine [31] were associated with poorer cognitive function, while spermidine [25], valine, and alpha-amino adipic acid [43] showed associations with better cognitive function. In the large-scale metabolomic study, glycoprotein acetyl, glutamate, and ornithine were associated with the impaired cognitive ability [38]. Another large untargeted metabolomic study found that maltose and other pertinent metabolites were also associated with worsening global cognitive function [45].

Table 4

The association between brain/blood metabolites and cognitive function in patients with cognitive decline

	Cognitive function test/score	Methods	Associated metabolites	Interpretation	References
Brain
Normal controls (n = 14, mean age 83±11)Asymptomatic AD (n = 15, mean age 89±8)AD (n = 14, mean age 88±9)	The longitudinal cognitive performance or cognitive performance at the last visit	Untargeted (GC-MS, LC-MS)	(+) linoleic acid(+) linolenic(+) EPA(+) oleic acid(+) arachidonic acid(–) DHA	Decreases in fatty acids were associated with cognitive decline (except DHA) in AD patients.	[26]
Blood
Normal controls (n = 115, mean age 78±7)Incident AD (n = 92, mean age 80±7)	Cognitive performanceAttentionLanguageVisuospatial ability	Targeted (FIA-MS/MS for lipids and LC-MS/MS for amino acids and biogenic amines)	(–) SM C18:1(–) PC aa C40:6(–) SM C26:1(–) PC ae C40:1(–) LysoPC a C18:0(–) arginine(+) spermidine	Increases in SMs, PCs, LysoPCs, and arginine and a decrease in spermidine were associated with cognitive decline in AD patients.	[25]
Non-AD converter (n = 436, mean age = 81±7)AD converter (n = 85, mean age = 86±6)	Global cognition	Targeted (FIA-MS/MS for acylcarnitines and lipids and LC-MS/MS for amino acids and biogenic amines)	(+) decanoylcarnitine (C10)(+) pimelylcarnitine (C7-DC)(+) tetradecadienylcarnitine (C14:2)(–) Asparagine(–) PC ae C34:0	Increases in acylcarnitines and decreases in asparagine and PCs were associated with cognitive decline in AD patients.	[31]
Normal controls (n = 370, mean age 75±6)Early MCI (n = 284, mean age 71±8)Late MCI (n = 505, mean age 74±8)AD (n = 305, mean age 75±8)	ADAS-Cog13	Targeted (LC-MS/MS, UPLC-MS/MS)	secondary bile acid(+) GDCA(+) GLCAsecondary/primary bile acid ratio(+) DCA:CA(+) GDCA:CA(+) TDCA:CA(+) GLCA:CDCA	Increases in secondary bile acids were associated with cognitive decline in AD patients.	[40]
AD (n = 35, mean age 73±4) (no significant correlation in normal controls and MCI)	MMSE	Targeted (LC-MS/MS)	(+) Medium-chain acylcarnitines	Decreases in medium-chain acylcarnitines were associated with cognitive decline in AD patients.	[32]
Normal controls (n = 199, age 75 (72–78))MCI (n = 358, age 75 (70–80))AD (n = 175, age 76 (71–80))	ADAS-Cog13 score	Targeted (UPLC-MS/MS)	(+) C14:1(+) C16:1(+) SM C20:2(–) α-aminoadipic acid(–) valine	Increases in acylcarnitine, SMs, PCs, and decreases in α-aminoadipic acid and valine were associated with cognitive decline in AD patients.	[43]
Normal controls (n = 199, age 75 (72–78))MCI (n = 358, age 75 (70–80))AD (n = 175, age 76 (71–80))	ADASCog13 changes up to 5 years	Targeted (UPLC-MS/MS)	(+) PC ae C40:3,(+) PC ae C42:4(+) PC ae C44:4(+) SM (OH) C14:1(+) SM C16:0(+) SM C20:2(–) α-aminoadipic acid(–) valine	Increases in SPHs and PCs and decreases in α-aminoadipic acid and valine were associated with cognitive decline in AD patients.	[43]
11 cohort study (n = 5,658, age 49–74)	General cognitive ability	Targeted (NMR)	(+) free cholesterol in small HDL(+) DHA(–) glycoprotein acetyl(–) Glutamate(–) Ornithine	Increases in free cholesterol in small HDL and DHA and decreases in glycoprotein acetyl, glutamate, and ornithine were associated with cognitive decline in the general population.	[38]
Bogalusa Heart Study (BHS) (n = 1,177, mean age 48±5)	Global cognitive scoreDigit coding test score	Untargeted (UPLC-MS)	(–) Maltose(–) N acetyl-isoputreanine(–) 9-hydroxystearate(–) 7-methylguanine(–) unknown metabolite X-21840	Decreases in maltose, N acetyl-isoputreanine, 9-hydroxystearate, and 7-methylguanine were associated with cognitive decline.	[45]

AD, Alzheimer’s disease; ADAS-Cog13, Alzheimer’s Disease Assessment Scale–Cognitive Subscale 13 items; CA, cholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; DHA, docosahexaenoic acid; EPA, Eicosapentaenoic acid; FIA-MS/MS, flow-injection tandem mass spectrometry; GC-MS, gas chromatography-mass spectrometry; GDCA, glycodeoxycholic acid; GLCA, glycolithocholic acid; HDL, high-density lipoprotein; LC-MS, liquid chromatography-mass spectrometry; LC-MS/MS, liquid chromatography-tandem mass spectrometry; LysoPCs, Lysophosphatidylcholines; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; NMR, Nuclear magnetic resonance; PCs, phosphatidylcholine; SMs, sphingomyelin; TDCA, taurodeoxycholic acid; UPLC-MS, ultraperformance liquid chromatography-mass spectrometer; UPLC-MS/MS, ultraperformance liquid chromatography-tandem mass spectrometer.

METABOLIC ALTERATIONS AS PREDICTORS FOR AD AND DEMENTIA IN METABOLOMICS RESEARCH

Some of the metabolomic studies in humans were conducted with prospective design to explore thepredictive biomarkers for AD and dementia (Table 5) [25 , 46–48]. One study showed that a panel of twelve metabolites could separate MCI or AD from control healthy participants within two years of the follow-up period [46]. The elevated levels of some SMs also increased the risk of developing AD when compared with non-AD participants within four years of follow-up time [25]. The predictability of phospholipids and acylcarnitines depended on specific species of these molecules since they showed different directions among the specific types [25 , 47]. The untargeted metabolomics study with a follow-up time of around eight years found that the alterations of metabolites from the consumption of diets and some endogenous metabolites such as hormones or amino acids were associated with the risk of developing AD from a non-AD state [47]. The increase in a cholesterol ester of HDL and the decrease in branched-chain amino acids were associated with a higher risk of AD [48]. The alterations in these metabolites before the development of AD might be related to the pathophysiologic changes of several organ systems or diet consumption, which could lead to the progression of AD.

Table 5

Blood metabolites at the baseline as a predictor for cognitive impairment in patients

	Follow time (y)	Methods	Major Metabolic difference at baseline in AD converter		Interpretation	References
			Lipid	Amino acid, polyamine, and other metabolites
Normal controls (n = 68, mean age 82±3)MCI/AD converter (n = 28, mean age 80±4)	2.10	Targeted (FIA-MS/MS for lipid and LC-MS/MS for others)	↑ hydroxyhexadecadienylcarnitine (C16:2-OH)↑ 3-hydroxypalmitoleylcarnitine (C16:1-OH)↓ valerylcarnitine (C5)↑ LysoPC a C28:1↑ LysoPC a C17:0↓ PC aa C38:5	↑ histamine↓ aspartate↑ L-Arginine↑ asparagine↓ citrulline↑ nitrotyrosine	Changes in metabolites in lipid and amino acid pathways increased the risk of developing dementia or AD.	[46]
Non-AD converter (n = 115, mean age 78±7)AD converter (n = 92, mean age 80±7)	4.27	Targeted (FIA-MS/MS for lipids)	↑ SM C16:0↑ SM C16:1↑ SM (OH) C14:1↑ SM C18:1)↓ PC aa 38:4↑ PC ae C34:2	NA	Elevated SMs and alterations in PCs increase the risk of developing AD.	[25]
Non-AD converter (n = 436, mean age 81±7)AD converter (n = 85, mean age 86±6)	4.50	Targeted (FIA-MS/MS for lipids)	↓ decanoylcarnitine (C10)↓ pimelylcarnitine (C7-DC)↓ tetradecadienylcarnitine (C14:2)	NA	Decreased acylcarnitines increased the risk of developing AD.	[31]
Non-AD converter (control) (n = 209, mean age 76±4)AD converter (case) (n = 209, mean age 76±4)	8.50	Untargeted (LC-MS)	↓ LysoPC C18:3↓ undecanoylcarnitine↑ lauroylcarnitine↑ myristoylcarnitine	Exogenous substances↓ atractylgenin glucuronide (coffee)↓ cyclo(leucyl-propyl) (coffee)↑ caffeine↑ proline betaine (citrus Juice)↓ CMPFP (fish consumption)↓ cyclo(propyl-valyl) (coca consumption)Endogenous substances↓ cortisol↑ GDCA↑ glucose↑ creatinine↑ trimethyllysine↑ L-arginine	Alterations in metabolites related to diet were associated with AD development.	[47]
MCI (n = 366, mean age 75±7)AD converter (n = 185, mean age 75±7)	2.97	Targeted (FIA-MS/MS for lipids)	↑ SM C18:1↑ PC aa 38:4	NA	Increased SMs and PCs increased the risk of developing AD from MCI.	[25]
MCI (n = 538, mean age 72±8)AD converter (n = 251, mean age 74±7)	3.94	Targeted (LC-MS/MS, UPLC-MS/MS)	NA	primary bile acid↓ CAsecondary/primary bile acid ratio↑ GDCA: CA↑ TDCA: CA	Increases in secondary bile acids increased the risk of developing AD from MCI.	[40]
MCI (n = 91, mean age 72±5)AD converter (n = 52, mean age 71±6)	2.20 –2.30	Untargeted (UPLC-MS for lipids and GC-MS for others)	↑ 2,4-dihydroxybutanoic acid↑ PC (16:0/16:0)↑ unidentified carboxylic acid	↑ pyruvic acid↑ lactic acid↓ ribose-5-phosphate	Increased fatty acid, pyruvic acid, lactic acid, and decreased ribose-5-phosphate increased the risk of developing AD from MCI.	[35]
Normal controls (n = 24, mean age 66±9)Incident dementia (n = 12, mean age 65±9)	5.00	Targeted (GC-MS)	↑ docosapentaenoic acid↑ 3,4-dihydroxybutanoic acid	↑ uric acid	The increases in metabolites in the lipid pathway including docosapentaenoic acid and 3,4-dihydroxybutanoic acid, and the increase in uric acid increased the risk of developing dementia.	[39]
Dementia-free (n = 1,974, mean age 55±10)Incident dementia (n = 93, mean age 68±6)	15.60	Targeted (LC-MS)	NA	↑ plasma anthranilic acid↑ glutamic acid↓ taurine↓ hypoxanthine	Increases in glutamic acid, anthranilic acid, and decreases in taurine, and hypoxanthine increased the risk of developing dementia.	[49]
Total participant (n = 22,623, age 50–75)Incident dementia (n = 995)AD (n = 745)	246,698 person-years	Target (NMR, LC-MS/MS)	↑ cholesterol ester in large HDL↑ total cholesterol to total lipids ratio in very large VLDL (dementia only)↓ cholesterol in small VLDL (dementia only)↓ triglycerides to total lipids ratio in very large VLDL (dementia only)	↓ BCAAs (isoleucine, leucine, and valine)↓ creatinine (dementia only)	Decreases in BCAAs and an increase in cholesterol ester in large HDL increase the risk of developing dementia or AD. A decrease in creatinine and the perturbation in VLDL subfraction were related to dementia but not AD.	[48]

AD, Alzheimer’s disease; BCAAs, branched-chain amino acids; CA, cholic acid; CMPFP, carboxy-4-methyl-5-pentyl-2-furanpropanoic acid; FIA-MS/MS, flow-injection - tandem mass spectrometry; GC-MS, gas chromatography - mass spectrometry; GC-MS/MS, gas chromatography - tandem mass spectrometry; GDCA, glycodeoxycholic acid; HDL, high-density lipoprotein; HPLC-MS/MS, high-performance liquid chromatography - tandem mass spectrometry; LC-MS, liquid chromatography - mass spectrometry; LC-MS/MS, liquid chromatography - tandem mass spectrometry; LysoPCs, Lysophosphatidylcholines; MCI, mild cognitive impairment; NA, not available; NMR, Nuclear magnetic resonance; PCs, phosphatidylcholine; SMs, sphingomyelin; TDCA, taurodeoxycholic acid; UPLC-MS/MS, ultraperformance liquid chromatography - tandem mass spectrometer.

Patients with MCI have a higher risk of developing AD later in their life. Therefore, several studies focused on the alterations in metabolites in MCI as AD predictors [25, 40]. The sphingolipid SM C18:1 and the phospholipid PC aa 38:4 were associated with a higher risk of AD conversion from MCI [25]. Furthermore, the untargeted-metabolomic study showed that the elevation of 2,4-dihydroxybutanoic acid, PC (16:0/16:0), an unidentified carboxylic acid, lactic acid, and pyruvate, and the reduction of ribose-5-phosphate were observed in MCI patients who later developed AD [40]. The changes in these metabolites are associated with hypoxia, oxidative stress, and lipid membrane remodeling. Additionally, there is evidence that gut dysbiosis is involved in the progression to AD via the alteration of metabolites as indicated by a decrease in primary bile acid and increases in secondary bile acids [40]. In general, metabolomic studies reveal some pathophysiologies of AD that occur before the development of the symptoms.

Regardless of the causes of cognitive impairment, some studies demonstrated that the alterations in blood metabolites were also associated with the incidence of dementia [39 , 49]. The increases in docosapentaenoic acid, 3,4-dihydroxybutanoic acid, and uric acid were found in the patients who developed dementia after approximately five years of follow-up [39]. In another study, the alterations in blood metabolites including increased plasma anthranilic acid and glutamic acid, and decreased taurine and hypoxanthine were associated with the risk of dementia development in the next fifteen years [49]. In the large prospective cohort study, one VLDL- and one HDL- specific lipoprotein lipid subclass were associated with increased risk of dementia, while three branched-chain amino acids, creatinine, and two VLDL-specific lipoprotein lipid subclasses were associated with a lower risk of dementia [48]. The alterations in these metabolites might involve the pathophysiologies of cognitive impairment and possibly increase the risk of developing AD in the future. Therefore, the restoration of the levels of these metabolites might be one of the targets to reduce the chance of developing AD.

CONCLUSIONS AND FUTURE DIRECTIONS

The metabolic alterations in both animal and clinical studies are shown in Fig. 2 and the metabolic changes in AD models after received neuroprotective intervention are summarized in Fig. 3. Of these findings, the changes in lipid metabolites were most significant. The changes in lipid levels in the brain, especially of SM and PC, were related to the development of AD most probably because these lipids are the essential constituents of the cell membrane where Aβ processing occurs. The changes in the levels of glucose, TCA metabolites, acylcarnitine, and some fatty acid metabolites could reflect dysregulated energy utilization in the brain. The perturbation of amino acid levels and derivatives might be associated with impaired neurotransmitter function. These findings were closely related to AD status and brain pathologies of AD. Although the alterations in blood metabolites could not directly reflect the pathologic changes in the brain, they may be associated with systemic disease or health status which could precipitate the development of dementia and AD. The disturbances in blood amino acids and their derivatives, bile acids, and lipid metabolites in AD and dementia patients indicated that systemic oxidative stress, gut dysbiosis, and impaired lipid metabolism might involve alteration of blood metabolites which could directly or indirectly affect brain function and AD pathologies. The authors suggest that a prospective design of a further study into metabolomics and AD will give vital information regarding the metabolic changes which occur before disease manifestation. Findings from prospective studies will be useful in understanding the pathophysiology behind these metabolic changes and their association with both dementia and AD. Furthermore, details regarding the perturbations of metabolites found in prospective research could be used as identifiable novel biomarkers in making an early diagnosis of dementia and AD. That information could help the patients to receive earlier treatment as well as potentially increase the possibility of slowing down the progression of the cognitive impairment.

Fig. 2

The alterations in metabolites found in the brain (a) and the blood (b) in animal and human metabolomic-based studies. BCAAs, branched-chain amino acids; DHA, docosahexaenoic acid; DPA, docosapentaenoic acid; EAA, essential amino acids; FA, Fatty acid; GABA, gamma-Aminobutyric acid; GSH, glutathione; NAG, N-acetylglutamate; NEAA, non-essential amino acids; PC, Phosphatidylcholine; PE, Phosphatidylethanolamine; PUFA, Polyunsaturated fatty acid; SAM, S-adenosylmethionine; SDMA, symmetric dimethylarginine; SM, Sphingomyelin; TCA, tricarboxylic acid.

Fig. 3

The alterations in metabolites found in the blood and/or the brain from in vivo models of Alzheimer’s disease following neuroprotective interventions. (a) Male SAMP8 mice after J147 treatment, (b) Male SAMP8 mice after Fisetin treatment, (c) Male Khun ming mice injected with Aβ_1-42 into the hippocampus after GRg1 treatment, and (d) Male Khun ming mice injected with Aβ_1-42 into the hippocampus after GRb1 treatment. Aβ, amyloid-β; BCAAs, branched-chain amino acids; cAMP, cyclic adenosine monophosphate; LysoPCs, Lysophosphatidylcholines; PUFA, Polyunsaturated fatty acid; SAMP8, senescence-accelerated prone 8.

There are some points to be included in a selection of the focus of future research. Most of the focus needs to be on the changes in the lipid metabolites which are related to the metabolism of the cell membrane, which is important for the function of the central nervous system. In addition, the pathophysiology of cognitive impairment is partially linked with the alterations of neurotransmitters in the brain. However, the knowledge surrounding these is still limited and inconsistent, and this information needs further investigation. Additionally, only a few clinical studies have investigated the association of metabolic alterations in both the brain and blood. The studies presented in this review mostly investigated the brain or the blood separately. Since the generalcharacteristics of participants and the criteria for diagnosis or categorization in each study differ extensively, these limit the interpretation of metabolic disturbances in the brain and the blood among the studies. Some of the animal studies included both brain and blood metabolomics, but the results were still controversial. In addition, when interpreting the data, it is important to accept that the metabolites between animal and human studies could be different.

Footnotes

ACKNOWLEDGMENTS

This work was supported by Senior Research Scholar Grant from the National Research Council of Thailand (SCC.), the NSTDA Research Chair Grant from the National Science and Technology Development Agency Thailand (NC), and the Chiang Mai University Center of Excellence Award (NC).

Authors’ disclosures available online ().

References

Arvanitakis

, Shah

, Bennett

(2019) Diagnosis and management of dementia: Review. JAMA 322, 1589–1599.

Reitz

, Mayeux

(2014) Alzheimer disease: Epidemiology, diagnostic criteria, risk factors and biomarkers. Biochem Pharmacol 88, 640–651.

Serrano-Pozo

, Frosch

, Masliah

, Hyman

(2011) Neuropathological alterations in Alzheimer disease. Cold Spring Harb Perspect Med 1, a006189.

Mishur

, Rea

(2012) Applications of mass spectrometry to metabolomics and metabonomics: Detection of biomarkers of aging and of age-related diseases. Mass Spectrom Rev 31, 70–95.

Chrast

, Saher

, Nave

, Verheijen

(2011) Lipid metabolism in myelinating glial cells: Lessons from human inherited disorders and mouse models. J Lipid Res 52, 419–434.

Pan

, Nasaruddin

, Elliott

, McGuinness

, Passmore

, Kehoe

, Holscher

, McClean

, Graham

, Green

(2016) Alzheimer’s disease-like pathology has transient effects on the brain and blood metabolome. Neurobiol Aging 38, 151–163.

Pan

, Green

(2019) Temporal Effects of Neuron-specific beta-secretase 1 (BACE1) Knock-in on the mouse brain metabolome: Implications for Alzheimer’s disease. Neuroscience 397, 138–146.

Klosinski

, Yao

, Yin

, Fonteh

, Harrington

, Christensen

, Trushina

, Brinton

(2015) White matter lipids as a ketogenic fuel supply in aging female brain: Implications for Alzheimer’s disease. EBioMedicine 2, 1888–1904.

Shang

, Mishra

, Wang

, Desai

, Chen

, Mao

, Do

, Bernstein

, Trouard

, Brinton

(2020) Evidence in support of chromosomal sex influencing plasma based metabolome vs APOE genotype influencing brain metabolome profile in humanized APOE male and female mice. PLoS One 15, e0225392.

10.

, Zhou

, Li

, Liu

, Wang

, He

(2015) Protective effects of ginsenosides Rg1 and Rb1 on an Alzheimer’s disease mouse model: A metabolomics study. J Chromatogr B Analyt Technol Biomed Life Sci 985, 54–61.

11.

Zhao

, Ren

, Yamazaki

, Qiao

, Li

, Felton

, Mahmoudiandehkordi

, Kueider-Paisley

, Sonoustoun

, Arnold

, Shue

, Zheng

, Attrebi

, Martens

, Li

, Bastea

, Meneses

, Chen

, Thompson

, St John-Williams

, Tachibana

, Aikawa

, Oue

, Job

, Yamazaki

, Liu

C-C

, Storz

, Asmann

, Ertekin-Taner

, Kanekiyo

, Kaddurah-Daouk

, Bu

(2020) Alzheimer’s risk factors age, APOE genotype, and sex drive distinct molecular pathways. Neuron 106, 727–742.e726.

12.

Plucinska

, Crouch

, Koss

, Robinson

, Siebrecht

, Riedel

, Platt

(2014) Knock-in of human BACE1 cleaves murine APP and reiterates Alzheimer-like phenotypes. J Neurosci 34, 10710–10728.

13.

Volianskis

, Køstner

, Mølgaard

, Hass

, Jensen

(2010) Episodic memory deficits are not related to altered glutamatergic synaptic transmission and plasticity in the CA1 hippocampus of the APPswe/PS1ΔE9-deleted transgenic mice model of β-amyloidosis. Neurobiol Aging 31, 1173–1187.

14.

Chene

, Beiser

, Au

, Preis

, Wolf

, Dufouil

, Seshadri

(2015) Gender and incidence of dementia in the Framingham Heart Study from mid-adult life. Alzheimers Dement 11, 310–320.

15.

Haughey

, Bandaru

, Bae

, Mattson

(2010) Roles for dysfunctional sphingolipid metabolism in Alzheimer’s disease neuropathogenesis. Biochim Biophys Acta 1801, 878–886.

16.

Raichle

, Gusnard

(2002) Appraising the brain’s energy budget. Proc Natl Acad Sci U S A 99, 10237–10239.

17.

Schonfeld

, Reiser

(2013) Why does brain metabolism not favor burning of fatty acids to provide energy? Reflections on disadvantages of the use of free fatty acids as fuel for brain. J Cereb Blood Flow Metab 33, 1493–1499.

18.

Nuriel

, Angulo

, Khan

, Ashok

, Chen

, Figueroa

, Emrani

, Liu

, Herman

, Barrett

, Savage

, Buitrago

, Cepeda-Prado

, Fung

, Goldberg

, Gross

, Hussaini

, Moreno

, Small

, Duff

(2017) Neuronal hyperactivity due to loss of inhibitory tone in APOE4 mice lacking Alzheimer’s disease-like pathology. Nat Commun 8, 1464.

19.

Currais

, Farrokhi

, Dargusch

, Armando

, Quehenberger

, Schubert

, Maher

(2018) Fisetin reduces the impact of aging on behavior and physiology in the rapidly aging SAMP8 mouse. J Gerontol A Biol Sci Med Sci 73, 299–307.

20.

Currais

, Goldberg

, Farrokhi

, Chang

, Prior

, Dargusch

, Daugherty

, Armando

, Quehenberger

, Maher

, Schubert

(2015) A comprehensive multiomics approach toward understanding the relationship between aging and dementia. Aging (Albany NY) 7, 937–955.

21.

Kurbat

, Lelevich

(2009) Metabolism of amino acids in the brain. Neurochem J 3, 23–28.

22.

Lenis

, Elmetwally

, Maldonado-Estrada

, Bazer

(2017) Physiological importance of polyamines. Zygote 25, 244–255.

23.

Chen

, Prior

, Dargusch

, Roberts

, Riek

, Eichmann

, Chiruta

, Akaishi

, Abe

, Maher

, Schubert

(2011) A novel neurotrophic drug for cognitive enhancement and Alzheimer’s disease. PLoS One 6, e27865.

24.

Khan

, Syed

, Ahmad

, Mukhtar

(2013) Fisetin: A dietary antioxidant for healthromotion. Antioxid Redox Signal 19, 151–162.

25.

Varma

, Oommen

, Varma

, Casanova

, An

, Andrews

, O’Brien

, Pletnikova

, Troncoso

, Toledo

, Baillie

, Arnold

, Kastenmueller

, Nho

, Doraiswamy

, Saykin

, Kaddurah-Daouk

, Legido-Quigley

, Thambisetty

(2018) Brain and blood metabolite signatures of pathology and progression in Alzheimer disease: A targeted metabolomics study. PLoS Med 15, e1002482.

26.

Snowden

, Ebshiana

, Hye

, An

, Pletnikova

, O’Brien

, Troncoso

, Legido-Quigley

, Thambisetty

(2017) Association between fatty acid metabolism in the brain and Alzheimer disease neuropathology and cognitive performance: A nontargeted metabolomic study. PLoS Med 14, e1002266.

27.

, Varma

, Casanova

, Dammer

, Pletnikova

, Chia

, Egan

, Ferrucci

, Troncoso

, Levey

, Lah

, Seyfried

, Legido-Quigley

, O’Brien

, Thambisetty

(2018) Evidence for brain glucose dysregulation in Alzheimer’s disease. Alzheimers Dement 14, 318–329.

28.

Mahajan

, Varma

, Griswold

, Blackshear

, An

, Oommen

, Varma

, Troncoso

, Pletnikova

, O’Brien

, Hohman

, Legido-Quigley

, Thambisetty

(2020) Dysregulation of multiple metabolic networks related to brain transmethylation and polyamine pathways in Alzheimer disease: A targeted metabolomic and transcriptomic study. PLoS Med 17, e1003012.

29.

Motsinger-Reif

, Zhu

, Kling

, Matson

, Sharma

, Fiehn

, Reif

, Appleby

, Doraiswamy

, Trojanowski

, Kaddurah-Daouk

, Arnold

(2013) Comparing metabolomic and pathologic biomarkers alone and in combination for discriminating Alzheimer’s disease from normal cognitive aging. Acta Neuropathol Commun 1, 28.

30.

Muguruma

, Tsutsui

, Noda

, Akatsu

, Inoue

(2018) Widely targeted metabolomics of Alzheimer’s disease postmortem cerebrospinal fluid based on 9-fluorenylmethyl chloroformate derivatized ultra-high performance liquid chromatography tandem mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci 1091, 53–66.

31.

Huo

, Yu

, Yang

, Zhu

, Bennett

, Zhao

(2020) Brain and blood metabolome for Alzheimer’s dementia: Findings from a targeted metabolomics analysis. Neurobiol Aging 86, 123–133.

32.

Ciavardelli

, Piras

, Consalvo

, Rossi

, Zucchelli

, Di Ilio

, Frazzini

, Caltagirone

, Spalletta

, Sensi

(2016) Medium-chain plasma acylcarnitines, ketone levels, cognition, and gray matter volumes in healthy elderly, mildly cognitively impaired, or Alzheimer’s disease subjects. Neurobiol Aging 43, 1–12.

33.

Weng

, Huang

, Tang

, Cheng

, Chen

(2019) The differences of serum metabolites between patients with early-stage Alzheimer’s disease and mild cognitive impairment. Front Neurol 10, 1223.

34.

Lin

, Huang

, Lin

, Yen

, Kuo

(2019) A metabolomic approach to identifying biomarkers in blood of Alzheimer’s disease. Ann Clin Transl Neurol 6, 537–545.

35.

Oresic

, Hyotylainen

, Herukka

, Sysi-Aho

, Mattila

, Seppanan-Laakso

, Julkunen

, Gopalacharyulu

, Hallikainen

, Koikkalainen

, Kivipelto

, Helisalmi

, Lotjonen

, Soininen

(2011) Metabolome in progression to Alzheimer’s disease. Transl Psychiatry 1, e57.

36.

Savica

, Murray

, Persson

, Kantarci

, Parisi

, Dickson

, Petersen

, Ferman

, Boeve

, Mielke

(2016) Plasma sphingolipid changes with autopsy-confirmed Lewy Body or Alzheimer’s pathology. Alzheimers Dement (Amst) 3, 43–50.

37.

Pena-Bautista

, Roca

, Hervas

, Cuevas

, Lopez-Cuevas

, Vento

, Baquero

, Garcia-Blanco

, Chafer-Pericas

(2019) Plasma metabolomics in early Alzheimer’s disease patients diagnosed with amyloid biomarker. J Proteomics 200, 144–152.

38.

van der Lee

, Teunissen

, Pool

, Shipley

, Teumer

, Chouraki

, Melo van Lent

, Tynkkynen

, Fischer

, Hernesniemi

, Haller

, Singh-Manoux

, Verhoeven

, Willemsen

, de Leeuw

, Wagner

, van Dongen

, Hertel

, Budde

, Willems van Dijk

, Weinhold

, Ikram

, Pietzner

, Perola

, Wagner

, Friedrich

, Slagboom

, Scheltens

, Yang

, Gertzen

, Egert

, Li

, Hankemeier

, van Beijsterveldt

CEM

, Vasan

, Maier

, Peeters

CFW

, Jorgen Grabe

, Ramirez

, Seshadri

, Metspalu

, Kivimaki

, Salomaa

, Demirkan

, Boomsma

, van der Flier

, Amin

, van Duijn

(2018) Circulating metabolites and general cognitive ability and dementia: Evidence from 11 cohort studies. Alzheimers Dement 14, 707–722.

39.

Mousavi

, Jonsson

, Antti

, Adolfsson

, Nordin

, Bergdahl

, Eriksson

, Moritz

, Nilsson

, Nyberg

(2014) Serum metabolomic biomarkers of dementia. Dement Geriatr Cogn Dis Extra 4, 252–262.

40.

MahmoudianDehkordi

, Arnold

, Nho

, Ahmad

, Jia

, Xie

, Louie

, Kueider-Paisley

, Moseley

, Thompson

, St John Williams

, Tenenbaum

, Blach

, Baillie

, Han

, Bhattacharyya

, Toledo

, Schafferer

, Klein

, Koal

, Risacher

, Kling

, Motsinger-Reif

, Rotroff

, Jack

, Hankemeier

, Bennett

, De Jager

, Trojanowski

, Shaw

, Weiner

, Doraiswamy

, van Duijn

, Saykin

, Kastenmuller

, Kaddurah-Daouk

, Alzheimer’s Disease Neuroimaging Initiative, the Alzheimer Disease Metabolomics Consortium (2019) Altered bile acid profile associates with cognitive impairment in Alzheimer’s disease-An emerging role for gut microbiome. Alzheimers Dement 15, 76–92.

41.

Luca

, Chattipakorn

, Sriwichaiin

, Luca

(2020) Cognitive-behavioural correlates of dysbiosis: A review. Int J Mol Sci 21, 4843.

42.

Graham

, Chevallier

, Elliott

, Holscher

, Johnston

, McGuinness

, Kehoe

, Passmore

, Green

(2015) Untargeted metabolomic analysis of human plasma indicates differentially affected polyamine and L-arginine metabolism in mild cognitive impairment subjects converting to Alzheimer’s disease. PLoS One 10, e0119452.

43.

Toledo

, Arnold

, Kastenmuller

, Chang

, Baillie

, Han

, Thambisetty

, Tenenbaum

, Suhre

, Thompson

, John-Williams

, MahmoudianDehkordi

, Rotroff

, Jack

, Motsinger-Reif

, Risacher

, Blach

, Lucas

, Massaro

, Louie

, Zhu

, Dallmann

, Klavins

, Koal

, Kim

, Nho

, Shen

, Casanova

, Varma

, Legido-Quigley

, Moseley

, Zhu

, Henrion

MYR

, van der Lee

, Harms

, Demirkan

, Hankemeier

, van Duijn

, Trojanowski

, Shaw

, Saykin

, Weiner

, Doraiswamy

, Kaddurah-Daouk

, Alzheimer’s Disease Neuroimaging Initiative, the Alzheimer Disease Metabolomics Consortium (2017) Metabolic network failures in Alzheimer’s disease: A biochemical road map. Alzheimers Dement 13, 965–984.

44.

Voyle

, Kim

, Proitsi

, Ashton

, Baird

, Bazenet

, Hye

, Westwood

, Chung

, Ward

, Rabinovici

, Lovestone

, Breen

, Legido-Quigley

, Dobson

, Kiddle

, Alzheimer’s Disease Neuroimaging Initiative (2016) Blood metabolite markers of neocortical amyloid-beta burden: Discovery and enrichment using candidate proteins. Transl Psychiatry 6, e719.

45.

Shi

, Bazzano

, He

, Gu

, Li

, Yaffe

, Kinchen

, Stuchlik

, Mi

, Nierenberg

, Razavi

, Kelly

(2019) Novel serum metabolites associate with cognition phenotypes among Bogalusa Heart Study participants. Aging (Albany NY) 11, 5124–5139.

46.

Mapstone

, Lin

, Nalls

, Cheema

, Singleton

, Fiandaca

, Federoff

(2017) What success can teach us about failure: The plasma metabolome of older adults with superior memory and lessons for Alzheimer’s disease. Neurobiol Aging 51, 148–155.

47.

Low

, Lefevre-Arbogast

, Gonzalez-Dominguez

, Urpi-Sarda

, Micheau

, Petera

, Centeno

, Durand

, Pujos-Guillot

, Korosi

, Lucassen

, Aigner

, Proust-Lima

, Hejblum

, Helmer

, Andres-Lacueva

, Thuret

, Samieri

, Manach

(2019) Diet-related metabolites associated with cognitive decline revealed by untargeted metabolomics in a prospective cohort. Mol Nutr Food Res 63, e1900177.

48.

Tynkkynen

, Chouraki

, van der Lee

, Hernesniemi

, Yang

, Li

, Beiser

, Larson

, Saaksjarvi

, Shipley

, Singh-Manoux

, Gerszten

, Wang

, Havulinna

, Wurtz

, Fischer

, Demirkan

, Ikram

, Amin

, Lehtimaki

, Kahonen

, Perola

, Metspalu

, Kangas

, Soininen

, Ala-Korpela

, Vasan

, Kivimaki

, van Duijn

, Seshadri

, Salomaa

(2018) Association of branched-chain amino acids and other circulating metabolites with risk of incident dementia and Alzheimer’s disease: A prospective study in eight cohorts. Alzheimers Dement 14, 723–733.

49.

Chouraki

, Preis

, Yang

, Beiser

, Li

, Larson

, Weinstein

, Wang

, Gerszten

, Vasan

, Seshadri

(2017) Association of amine biomarkers with incident dementia and Alzheimer’s disease in the Framingham Study. Alzheimers Dement 13, 1327–1336.