In this study, we examined the relationship between metabolites and brain imaging markers of AD/ADRD, and identified several metabolites, including secondary BA, BCAA and TMAO associated with these AD/ADRD markers in older Boston Area Puerto Rican adults.
Our results indicate a potential inverse association between trimethylamine-N-oxide (TMAO) on brain aging and hippocampal volume. TMAO is biosynthesized from choline, betaine, and L-carnitine by gut microbial metabolism. Blood TMAO increases after consumption of foods such as fish, eggs and red meat that are rich in L-carnitine and phosphatidylcholine, and recent studies have linked circulating TMAO to inflammation, type 2 diabetes, and cardiovascular disease (30, 31). The influence of TMAO on dementia risk is still under debate; however, TMAO can cross the blood-brain barrier, and high TMAO concentration can induce neuroinflammation by triggering proinflammatory signaling pathways via NF-kB and cytokines (31). Past studies have reported that AD and MCI patients had elevated TMAO in their CSF, suggesting that this metabolite plays a role in oxidative stress and may contribute to neurodegeneration (32). Our findings further corroborate the connection and potential negative role of TMAO in aging, inflammation, and AD/ADRD risk.
We identified inverse associations between two short-chain acyl carnitines, propionyl carnitine (C3) and isobutylryl carnitine (C4 and brain age, as well as phenylacetyl carnitine, intermediate metabolites of amino acid and fatty acid oxidation in mitochondria which have received attention for their potential use as biomarkers for different disorders, including schizophrenia (33, 34). Disruptions in metabolic profiles of acylcarnitines have been observed in Alzheimer’s disease and MCI(6). Propionyl carnitine (C3) is biosynthesized from propionyl-Co, a metabolite of branched-chain aminos acids (isoleucine, valine, threonine, methionine), odd-chain fatty acids, and side chains of cholesterol. Propionylcarnitine has been previously observed to be elevated in the blood of participants with obesity, type 2 diabetes (35), and heart failure (36). Similarly, isobutyryl carnitine (C4) is a product of the metabolic oxidation of valine and fatty acids. The accumulation of C3-, C4- and C5-acylcarnitines could be due to incomplete oxidation of fatty acids and branched-chain amino acids at the level of different acyl-CoA dehydrogenase enzyme (37). There is mounting evidence on the role of branched-chain amino acids in brain function and AD/ADRD risk (38). A recent study of metabolomic profiles in the Alzheimer’s Disease Neuroimaging Initiative cohort reported an association of propionylcarnitine with decreased amyloid-β accumulation and higher memory scores (6).
Our results support the previously observed link between bile acid metabolites and AD/ADRD. Table 2 shows elevated blood levels of six cholesterol derivatives (glycine and taurine conjugated forms) involved in secondary bile acid metabolism, and of glycochenodeoxycholate glucuronide, involved in primary bile acid metabolism, each of which were associated with negative cognitive trajectory. Bile acids are produced in the liver from cholesterol and are further processed by gut bacteria (4). Bile acids can cross the blood-brain barrier, bind to nuclear receptors, and initiate important signaling and regulatory responses. A dysregulated bile acid profile is associated with cognitive impairment and Alzheimer’s Disease (AD). A recent study comparing blood metabolites of Alzheimer’s Disease patients to those of cognitively normal older adults reported significantly lower serum concentration of primary bile acids (cholic acid CA) and increased concentrations of the bacterially produced, secondary BA, deoxycholic acid (DCA), and its glycine and taurine conjugated forms (4). In addition, several mutations in the genes involved in immune response were observed in AD profiles and showed associations with cholesterol metabolism (39) and proper neuronal myelination (40). A secondary bile salt, sodium taurocholate, showed a positive association with cognition in our study. Sodium taurocholate sulfate is a bile salt that forms micelles around insoluble lipids, such as cholesterol, for their transport. The sodium/taurocholate cotransporting peptide (NTCP) is one of the key basolateral transporter proteins that maintain bile salt homeostasis and is highly expressed in brain tissue (41).
Another metabolite identified in this study is p-cresol glucuronide, a gut microbial co-metabolite of amino acids phenylalanine and tyrosine, formed via p-cresol. While one recent study found that it promotes blood-brain barrier integrity in mice (42), other studies report this metabolite as a uremic toxin linked to cardiac-related mortality and chronic kidney disease (43). Moreover, a recent study of metabolic profiles of bladder cancer suggested p-cresol-glucuronide as a potential biomarker for bladder cancer detection (44). Our results show that p-cresol glucuronide is correlated to metabolites carrying the phenylacyl group, specifically phenylacetyl glutamate, phenylacetyl carnitine and phenylacetyl glutamine. The most studied of this group is a microbial derived metabolite phenylacetyl glutamine. Multiple studies suggest that phenylacetyl glutamine plays an important role in brain and other physiological processes, and that it could be a biomarker in acute ischemic stroke (45) and type 2 diabetes (46).
Several metabolites identified in this study have been previously linked with diabetes and metabolic syndrome, including BCAA and TMAO. Prior studies with the BPRHS cohort reported that the participants had higher prevalence of obesity and age-related conditions, such as type 2 diabetes and hypertension, than those in the general population. Growing evidence shows that diabetes predisposes to structural brain abnormalities and cognitive decline, leading to dementia (47). A recent BPRHS study involving brain MRI imaging reported a potential impact of comorbid diabetes and hypertension on accelerated brain aging and cognitive impairment, with overall declines in executive function and global cognitive function (16). Moreover, the smallest hippocampal volumes and larger brain age deviations were observed among the participants with type 2 diabetes and hypertension.
Several metabolites were significantly associated with AD/ADRD outcomes in our study, including secondary bile acids, TMAO, phenylacetylcarnitine, and p-cresol-glucuronide are microbially derived. There is a close relationship between the host plasma metabolome and gut microbiota. As biochemical nutrients such as proteins, lipids, carbohydrates, etc. are digested and processed in the gut, they are catabolized into host-derived and microbial metabolites. Both types of metabolites can cross into the bloodstream and induce biological responses that affect different processes including those of the central nervous system (31). For example, the levels of microbially derived p-cresol glucuronide and phenylacetylglutamine showed strong associations with certain bacterial species in the gut. Metabolomics reflects the effects of genetic make-up, lifestyle, and environmental factors on a person’s overall health and cognition.
Understanding how the timing of blood metabolites is correlated to the risk of developing a disease has gained much attention in recent years (48, 49). In our study, the blood metabolites were measured on average 12.7 years prior to the brain imaging and cognitive outcomes. Blood metabolite makeup can vary depending on one’s lifestyle and changes in diet and medication over time (49), a limitation of this study. In our previous, cross-sectional report (18) that considered metabolite and cognitive data at baseline in the BPRHS, the key metabolites associated with cognitive function, were somewhat different from the current study, and were primarily sugars. However, MRI was only collected at the wave 4 visit in the BPRHS, so we were not able to examine MRI outcomes cross sectionally. While there are several reports on the long-term biomarker-outcome associations, data on the long-term within-person stability of metabolomics are generally lacking (49, 50). However, between-person metabolite variability tends to be higher than within-person variability. A recent report suggests that within-person metabolite stability over 10 years is reasonable for many lipid and polar metabolites, but can vary by metabolite class (49). The most stable metabolites were nucleotides and analogs, diglyceride, plasmalogens, bile acid precursors, and cholesterol esters, whereas lysophospholipids and some cholesterolderivatives showed lower stability. Diet-related metabolites are likely to have lower within-person stability, e.g. triglycerides with unsaturated fatty acids tend to be more stable over time than triglycerides with saturated fatty acids.
In summary, this study of the metabolome and MRI markers of brain aging over 12 y of follow-up in the BPRHS, the first such study in Puerto Ricans, identified several metabolites, including isobutyrylcarnitine, propionylcarnitine, phenylacetylglutamine, phenylacetylcarnitine, p-cresol-glucuronide, and trimethylamine N-oxide (TMAO) as associated facster brain aging, and taurocholate sulfate, a bile salt, was associated with better brain aging. These findings are generally consistent with studies on metabolomics and AD/cognitive decline in other populations. Further research on the role of the metabolome in underserved populations, particularly Latinos, is necessary to understand the mechanisms of the observed associations.