Association of Trimethylamine N-Oxide with Normal Aging and Neurocognitive Disorders: A Narrative Review

Aging-related neurocognitive disorder (NCD) is a growing health concern. Trimethylamine-N-oxide (TMAO), a gut microbiota-derived metabolite from dietary precursors, might emerge as a promising biomarker of cognitive dysfunction within the context of brain aging and NCD. TMAO may increase among older adults, Alzheimer’s disease patients, and individuals with cognitive sequelae of stroke. Higher circulating TMAO would make them more vulnerable to age- and NCD-related cognitive decline, via mechanisms such as promoting neuroinflammation and oxidative stress, and reducing synaptic plasticity and function. However, these observations are contrary to the cognitive benefit reported for TMAO through its positive effects on blood–brain barrier integrity, as well as from the supplementation of TMAO precursors. Hence, current disputable evidence does not allow definite conclusions as to whether TMAO could serve as a critical target for cognitive health. This article provides a comprehensive overview of TMAO documented thus far on cognitive change due to aging and NCD.


Introduction
The increased pace of aging has been both a challenge and a triumph for public health. Globally, 1 in 6 people will be aged 60 years or above by 2030 [1]. Advanced age is one of the strongest risk factors for neurocognitive disorders (NCD) such as Alzheimer's disease (AD) and poststroke cognitive impairment (PSCI) [2]. Driven by the aging population, the number of individuals with dementia is doubling every 20 years, and even among dementia-free seniors, 1 in 5 have mild NCD syndromes [3]. Patients with NCD typically exhibit longitudinal declines in cognitive and functional abilities, posing an increased risk for higher health expenditure [4] and all-cause mortality [2]. Although age increases risk, there are marked individual differences in the vulnerability for older adults in terms of NCD. Indeed, the onset of these disorders could be prevented or delayed, given that a wide range of predictive features, such as bio-behavioral factors, psychosocial characteristics, and cardiovascular diseases, are potentially modifiable [5,6]. Among them, gut microbiota and their metabolic products, including trimethylamine-N-oxide (TMAO), have recently emerged as a promising disease modifier and might play a putative role in aging and the development and progression of NCD [7][8][9].
TMAO is synthesized from the oxidation of trimethylamine (TMA) by hepatic flavin monooxygenases (FMO), and TMA is a microbiota-derived metabolite, generated in the gut from dietary precursors, mainly choline, L-carnitine, and betaine [10]. The TMAO level is affected by dietary intake, renal clearance, and host and microbial enzymes [11,12]. In contrast with a positive correlation observed between the dietary intake of TMAO precursors and cognitive function [13][14][15], elevated TMAO levels are detected in healthy aging [16] and associated with the pathogenesis of several diseases-for example, metabolic abnormalities, autoimmune disorders, colon cancer, and most prominently, atherosclerotic diseases [17,18]. A circulating TMAO level in the range of 1.5 µM to 10.5 µM has been shown to have a dose-dependent relationship with incident cardiovascular risk [19], the exposure of which serves as a vital target for cognitive health [20]. On top of that, TMAO itself contributes to cognitive decline both with aging [21] and NCD [22], suggesting an effect of higher TMAO levels on NCD per se, as well as their risk factors. Evidence shows that TMAO may be involved in the processes of brain aging and cognitive impairment via promoting neuroinflammation and oxidative stress, as well as reducing synaptic plasticity and function [23]. Despite this, opposing findings that TMAO had positive effects upon the blood-brain barrier (BBB) integrity and murine cognitive function in response to inflammatory challenge [24] illuminated the complex nature of the relationship that remains to be elucidated.
To date, whether TMAO could play a critical role in the pathophysiology of cognitive impairment due to aging and NCD, thus affecting disease development, or whether it could be a mediator contributing to disease progression, or is just a consequential marker, is still unclear. Therefore, the present review aims to summarize current evidence regarding the potential relationship of TMAO with normal cognitive aging and NCD.

Normal Cognitive Aging
The TMAO level may rise during the normal aging process [21,25] (Table 1). Healthy individuals aged above 65 years had significantly increased plasma TMAO levels, with a mean plasma concentration of 9.8 µM, in comparison to 4.4 µM in adults aged 45 to 64 years, and 2.8 µM in adults aged 18 to 44 years, revealing that TMAO was positively related to age (r 2 = 0.161, p < 0.001) [25]. Moreover, brain TMAO levels were also higher in old vs. young mice, and were highly correlated to the levels in circulation [21], indicating a direct effect of TMAO on the brain and cognitive function.
Indeed, in healthy middle-aged to older adults, circulating TMAO levels inversely predicted working memory and fluid cognition independent of traditional risk factors [21]. In mice, 16-week treatment of TMAO at a concentration of 1.5% could induce and aggravate brain aging and aging-related cognitive dysfunction as a result of neuron senescence, and the underlying mechanism would be the mitochondrial impairments driven by oxidative stress and the reduced expression of synaptic plasticity-related proteins by inhibiting the mammalian target of rapamycin (mTOR) signaling pathway [25]. Likewise, preexisting higher circulating TMAO may sensitize sevoflurane-induced cognitive impairment in aged rats, probably via downregulating antioxidant enzyme methionine sulfoxide reductase A in the hippocampus, then leading to microglia-mediated neuroinflammation [26]. Furthermore, TMAO would induce aging-like cognitive impairments in young animals, as TMAO-supplemented young adult mice performed worse on the novel object recognition test compared to the controls, with higher concentrations of pro-inflammatory cytokines and the reactive astrocyte marker, suggesting that TMAO might mediate cognitive aging by inducing neuroinflammation and astrocyte activation [21] (Table 1). Therefore, the potential of TMAO as a prevention and/or treatment target for cognitive declines in aging has been exploited ( Table 1). The aging mice demonstrating deficits in memory and cognitive function revealed an improvement in cognition after Fructus Ligustri Lucidi (i.e., the ripe fruit of Ligustrum lucidum Ait) treatment, possibly by lowering oxidative stress subsequent to decreased circulating TMAO levels via the altered gut microbiota, characterized as a reduction in Bifidobacterium and Lactobacillus, and an increase of the Sutterella, Unclassified_Clostridiales, Corpococcus, and Clostridium, among others [27]. However, nutritional intake of TMA precursors might have cognitive protection capacities. For instance, a randomized clinical trial has shown that the supplementation of 2 g L-carnitine taken orally once a day for six months significantly improved the cognitive function in subjects above 100 years of age, showing significant improvements in the mini-mental state examination (MMSE) score (4.10 compared with 0.60) when compared to the placebo group [28].

Alzheimer's Disease
In 2017, Del Rio et al. provided the first evidence that TMAO can be detected in the human cerebral spinal fluid (CSF) [29] ( Table 2). The presence of TMAO in the CSF might be hepatic-derived, as TMAO could cross the BBB [30], and high amounts of circulating TMAO might be linked to BBB disruption in vivo [31], yet de novo synthesis of TMAO could also be possible since FMO3 has been detected in the adult brain [32]. The CSF levels of TMAO in AD patients were elevated compared to cognitively unimpaired individuals [33] but were similar to that in patients with mild cognitive impairment (MCI) [33], non-ADrelated dementia, or other neurological disorders [29]. Consistently, urine TMAO levels were comparable between MCI and AD patients, and both were higher than cognitively healthy controls (mean TMAO levels in the urine, cognitively healthy controls: 10.2 µM, MCI: 19.9 µM, AD: 18.9 µM) [3].
The intriguing role of TMAO in AD etiology was indicated in a 2016 study using a network-based ranking algorithm [22]. Among the known AD-associated microbial metabolites in humans, TMAO ranked as the top one that shared significant genetic commonality with AD. There were nine co-regulated genetic pathways, including pathways related to "AD", "axon guidance", "immune systems", "neuron signaling", and "lipid and protein metabolism" [22]. Interestingly, in vitro studies demonstrated that TMAO affected amyloid-β (Aβ) conformation and facilitated Aβ aggregation [34,35]; moreover, the reduction of plasma TMAO levels, either by treatment of 3,3-Dimethyl-1-butanol (DMB) or a combination of Lactobacillus plantarum and memantine, led to significantly decreased concentration of Aβ42, Aβ40, and Aβ deposition in the hippocampus of APP/PS1 mice [36,37]. However, CSF TMAO levels were not significantly correlated with Aβ42/Aβ40, but with CSF phosphorylated tau (p-tau) and p-tau/Aβ42 [33], the aggregation of which shown to be enhanced by TMAO by using in vitro models [38,39] (Table 2).
An association of TMAO with neurodegeneration has also been displayed alongside the pathological profile, that the CSF TMAO levels were more closely related to the biomarker of axonal injury (CSF total tau and neurofilament light chain protein) than the biomarker of dendritic degeneration (CSF neurogranin) [33]. Meanwhile, circulating TMAO levels have been related to hippocampal neuroinflammation. Downregulation of TMAO levels in plasma could alleviate the neuroinflammatory state of AD model mice, as indicated by the significant reductions in plasma clusterin, together with IL-2, IL-17, and TNF-α levels in the hippocampus [36,37]. Furthermore, TMAO impaired synaptic plasticity in the form of reduced long-term potentiation (LTP) through the endoplasmic reticulum (ER) stress-mediated PERK signaling pathway [40] ( Table 2).
The increased plasma TMAO in high-fat feeding dementia-prone (3xtg) mice [41] and choline-supplemented mice [37] also correlated with cognitive impairment. Lowering TMAO by DMB treatment or Lactobacillus plantarum supplementation protected hippocampal neuronal integrity and plasticity [37] and ameliorated LTP and cognitive decline in AD transgenic mice [36,37]. Nonetheless, these associations do not reflect causality. Accord-ing to a recent bidirectional Mendelian randomization study, TMAO and its precursors, including choline, carnitine, and betaine, did not have a causal effect on the risk of AD [42] ( Table 2).

Poststroke Cognitive Impairment
The predictive role of elevated TMAO has been evaluated in the development of PSCI. For patients with first-ever ischemic stroke admitted <7 days of symptom onset, higher plasma TMAO within 24 h of admission may increase the likelihood of PSCI assessed by the MMSE score ≤ 26 at 1 year after stroke (highest TMAO quartile > 7.4 µM vs. lowest TMAO quartile < 3.9 µM: adjusted odds ratio, aOR, 3.304; 95% confidence intervals, 95% CI, 1.335-8.178; p = 0.010) [44]. The admission plasma TMAO levels were also higher in patients suffering from a minor stroke in the past 2 weeks who had PSCI defined as < 22 points on the Montreal Cognition Assessment (MoCA) at 6-12 months after stroke onset than those without PSCI (median plasma TMAO: 4.6 µM vs. 3.2 µM; p ≤ 0.001), but there were no significant differences in circulating levels of TMAO precursors such as L-carnitine and choline [45] (Table 3).
Nevertheless, clinical findings investigating the relationship between PSCI and TMAO have been inconsistent. Zhong et al. showed that the baseline plasma TMAO level within 72 h of stroke onset was only associated with MMSE-defined PSCI (i.e., MMSE score < 27), with an aOR of 1.33 (95% CI, 1.04-1.72) for each 1-SD increment of TMAO, but there was no significant relationship when PSCI was defined according to MoCA score < 25, while notably, an inverse dose-response relationship was observed for TMAO precursors (choline and betaine) with either MMSE-defined or MoCA-defined PSCI [46]. A possible explanation for this discrepancy, besides the different definitions of PSCI, arises from variations in the timing of blood collection as TMAO levels were observed to elevate within 24 h after symptom onset, but then decreased in the following week, although it increased again after three months [47,48] (Table 3).
Experimental evidence is scarce. In mice undergoing repeated cerebral ischemiareperfusion injury, plasma TMAO levels increased significantly, and were associated with cognitive and LTP decline, decreased functional connectivity, reduced neuronal plasticity, and dendritic spine density, along with higher levels of pro-inflammatory cytokines of IL-1β, IL-6, and TNF-α in the hippocampus after ischemia [49]. Administration of baicalin (50 and 100 mg/kg), a flavonoid from Scutellaria baicalensis with neuroprotective properties, could restore normal plasma levels of TMAO by modifying the composition of the intestinal microbiota, subsequently improve cognition and attenuate the neuropathology related to cerebral ischemia-reperfusion injury [49] (Table 3).

Cognitive Frailty in Cardiovascular Disease
One study explored TMAO and cognitive frailty in older adults with cardiovascular disease [50]. In this cross-sectional setting, patients aged 65 years or older who had cardiovascular disease were assessed in terms of cognitive frailty, as determined by the simultaneous presence of physical frailty (three or more of the five conditions: weight loss, exhaustion, low activity, weakness, and slowness) and an MMSE score of ≤25; consequently, an independent association was observed between each 2-unit increase in TMAO and cognitive frailty (aOR, 1.21; 95% CI, 1.01-1.45; p = 0.04) [50] (Table 3).

Parkinson's Disease Dementia
It is known that the pathology of Parkinson's disease (PD) is linked to α-synuclein misfolding and aggregation, whereas TMAO has been shown to shift α-synuclein structures toward a more compact protein dimension in vitro [51]. As expected, PD patients with normal cognition had higher plasma TMAO levels than healthy controls (6.33 ± 0.56 vs. 3.76 ± 0.38 µM; p = 0.020) [52]. However, when comparing PD with normal cognition to patients with Parkinson's disease dementia (PDD, diagnosed by an MMSE score of 25 or less combined with any impairment in the eight instrumental activities), no difference was detected in plasma TMAO levels (6.33 ± 0.56 vs. 10.61 ± 4.53 µM; p = 0.220) [52]. Neither was the TMAO cut-point of 4.88 µM (obtained using the Youden index) distinguishable for the longitudinal cognitive decline based on a sustained decrease of at least 2 points in the MMSE score (p = 0.191). Yet there was a multivariable-adjusted trend that higher baseline TMAO levels predicted the risk of cognitive deterioration after a mean period of 4.3 ± 2.2 years (adjusted hazard ratio, aHR, 1.001; 95% CI, 1.000-1.002; p = 0.091) [52] ( Table 4). However, it is still too early to determine if TMAO is a risk factor for PDD. There were conflicting findings on the association between TMAO and cognitive progression in PD patients in another longitudinal study with a mean follow-up duration of 2.75 ± 0.60 years [53]. In this study, PDD was diagnosed under the clinical diagnostic criteria, and briefly, cognitive impairment in at least two cognitive domains with abnormality in activities of daily living was required. In patients with drug-naïve, early-stage PD, the plasma TMAO level was not associated with the risk for PDD conversion (p = 0.488), but a lower baseline plasma TMAO level (<6.92 µM) tended to be independently associated with a lower risk for dementia conversion (aHR, 7.565; 95% CI, 1.004-57.019; p = 0.050). After removing cases with plasma TMAO levels that could be considered outliers, the conclusion was the same that higher baseline plasma TMAO levels tended to be associated with a lower risk for PDD conversion (aHR, 0.182; 95% CI, 0.028−1.174; p = 0.073) [53] (Table 4).

Current Gaps
Recent findings revealing the impact of aging on TMAO and the association between TMAO and the development and progression of NCD are intriguing (Tables 1-5), although many questions need to be solved. Most importantly, whether TMAO is a friend or foe of cognition is still uncertain. Observations are contradictory that cognition could be improved both after reduction of TMAO [27,36,37] and after supplementation of TMAO precursors [13][14][15]28]. Another concern arises from the evidence that TMAO has antiinflammatory properties through beneficially regulating BBB integrity [24], but to date, most investigations have concluded the opposite; thus, the precise mechanisms of the crosstalk between TMAO and inflammation remain open. In clinical settings, it is consistent that TMAO increases with aging and in individuals with AD, but this feature might not be useful in the differential diagnosis or progression prediction for AD [29,33]. Additionally, it is yet to be determined whether TMAO affects cognitive aggravation in patients suffering from other types of NCD, such as PSCI and PDD. Further studies are required to explore if the baseline level of plasma TMAO is able to predict PSCI in stroke patients, and to rectify the conflicting effects of TMAO on cognitive decline in PD patients, better taking into account the sampling time of TMAO and the variations in defining criteria of cognitive progression.

Conclusions
Current evidence has demonstrated a link between TMAO and cognitive change due to aging and NCD. However, the nature of this link remains vague, because there is clear evidence not only for TMAO to influence the pathology underlying aging and NCD but also for these conditions to elevate the circulating TMAO level. We hypothesize that this truly is a bidirectional interaction; for example, aging or NCD itself may induce changes in plasma TMAO concentrations, and these changes may then increase the susceptibility to cognitive impairment. Studies to date have outlined potential mechanistic roles for such communication, often acting via neuroinflammation. However, it seems highly unlikely that TMAO is a driving force or the only player involved in determining variations in cognitive status in the context of aging and NCD. Looking forward, studies using preclinical models are required to assess the interaction between cognition and TMAO as well as its precursors, the producing enzyme FMO3, in different stages of NCD. Meanwhile, longitudinal clinical studies that dynamically assess the changes in TMAO levels with short-term and long-term cognitive trajectory would help to resolve uncertainty about the diagnostic potential of TMAO for cognitive aging and NCD. In short, the bulk of the evidence suggests that at least TMAO is increased with aging and AD, but more mechanistic understandings are needed to determine with certainty the effects of TMAO on aging and NCD-related cognitive impairment and progression.