Gut Microbiota-Derived Metabolites and Cardiovascular Disease Risk: A Systematic Review of Prospective Cohort Studies

Gut microbiota-derived metabolites have recently attracted considerable attention due to their role in host-microbial crosstalk and their link with cardiovascular health. The MEDLINE-PubMed and Elsevier’s Scopus databases were searched up to June 2022 for studies evaluating the association of baseline circulating levels of trimethylamine N-oxide (TMAO), secondary bile acids, short-chain fatty acids (SCFAs), branched-chain amino acids (BCAAs), tryptophan and indole derivatives, with risk of cardiovascular disease (CVD). A total of twenty-one studies were included in the systematic review after evaluating 1210 non-duplicate records. There were nineteen of the twenty-one studies that were cohort studies and two studies had a nested case–control design. All of the included studies were of high quality according to the “Newcastle–Ottawa Scale”. TMAO was positively associated with adverse cardiovascular events and CVD/all-cause mortality in some, but not all of the included studies. Bile acids were associated with atrial fibrillation and CVD/all-cause mortality, but not with CVD. Positive associations were found between BCAAs and CVD, and between indole derivatives and major adverse cardiovascular events, while a negative association was reported between tryptophan and all-cause mortality. No studies examining the relationship between SCFAs and CVD risk were identified. Evidence from prospective studies included in the systematic review supports a role of microbial metabolites in CVD.


Introduction
Cardiovascular disease (CVD) remains a major public health issue [1]. The progress in its prevention and management depends on a better understanding of the mechanisms underlying disease development. Identification of circulating biomarkers with prognostic value may help to both identify pathophysiological processes relevant to CVD development and improve preventive cardiovascular risk reduction efforts [2]. Recent development of omics technologies has improved biomarker discovery, leading to the identification of a number of disease-associated circulating molecules [3]. Metabolomics has considerably The initial search yielded 1210 non-duplicate records which were screened and 1137 records were removed based on title and abstract, leaving 73 reports for full-text examination ( Figure 1). We excluded 52 reports for reasons indicated in Figure 1 and finally 21 studies were deemed eligible for inclusion in the present systematic review. Of them, 19 included prospective cohorts [18][19][20]22,[24][25][26][27][28][29][30][31]36,[38][39][40][41], one study included one nested case-control study [23] and one study included two nested case-control studies [21]. All of the included studies were of high quality with a mean of 8.9 points in the Newcastle-Ottawa scale (range: [7][8][9] (Tables S2 and S3).
With regards to bile acids, Cheng and colleagues performed metabolite profiling (LC-MS) on plasma samples collected from 2327 participants of the Framingham Offspring Study who underwent a routine examination between 1991 and 1995 and were followed up to 20 years [28]. During this period, 358 individuals developed CVD. Of the 217 metabolites analyzed, three secondary bile acids (glycocholate, glycodeoxycholates, deoxycholates) were not significantly associated with CVD [28].
Another secondary bile acid, the deoxycholic acid was also not significantly associated with risk of CVD in 3147 patients with chronic kidney disease after the inclusion of many potential confounders in the analyses conducted by Frazier et al. [32].
On the other hand, in a large cohort of women free of CVD (Women's Health Study) 27,041 subjects who were followed over a mean 18.6 years elevated baseline plasma levels of total BCAAs assessed by NMR were found associated with higher risk of CVD (per 1SD increase; HR 1.14 [95% CI 1.08-1.18]) [39].

MACE
The MACE components varied by studies and a detailed description can been seen in Table 1.
In 1463 patients with T2D, associations between TMAO plasma concentrations and MACE were investigated by Croyal et al. [22]. During a follow-up of 7.1 years, 403 MACE cases were ascertained. In multivariate Cox regression models (covariates: demographics, personal history of MI, renal function markers, NT-proBNP), compared with the first quartile, those patients in the fourth quartile of TMAO concentrations were at increased risk for MACE (HR 1.29 [95% CI 1.02-1.64]) [22].
In another prospective analysis of plasma TMAO conducted in 262 symptomatic PAD patients by Roncal et al., during a mean period of 4 years, 135 cases were recorded. Multivariable Cox regression analyses (covariates: demographics, CRP, smoking, T2D, hypertension, dyslipidemia, HDL-C, renal function) did not reveal significant associations between one log unit increase in TMAO concentrations and risk of MACE [38].
In a study by Fan et al., 147 patients with CKD stage 1-5 were followed over a 3-year period, and 47 of them developed MACE [31]. Elevated levels of serum indoxyl sulfate were associated with an increased risk of MACE even after adjustment for demographics, traditional CVD risk factors, medications, renal function markers (HR 1.45 [95% CI 1.02-2.06]) [31].

CAD, Major Vascular Event, AF, HF, MI, Stroke, Coronary Revascularization
Liu et al. analyzed data from two cohorts, the Nurses' Health Study II and the Health Professionals Follow-up Study, 1216 participants free of CVD and T2D were followed and 608 developed CAD. TMAO was not independently associated with CAD risk in pooled analysis [23].
In another recent study by Papandreou, C et al., associations of plasma TMAO and its precursors with incident AF and HF were examined among 1879 subjects at high CV risk [21]. A total of two nested case-control studies were conducted within the PREDIMED study and after a mean follow-up for about 10 years, 512 AF and 334 HF incident cases were ascertained. Contrary to choline, betaine and dimethylglycine, TMAO was not associated with AF and HF [21].
TMAO was also not significantly associated with MI in a cohort of 1726 patients with suspected functionally relevant CAD [25].
Contrary to these findings, a more recent study by Chen et al. revealed that TMAO levels above the median were associated with higher risk of a major vascular event defined as a composite of transient ischemic attack, recurrent ischemic stroke, and MI in 291 patients with ischemic stroke [26].
In the Atherosclerosis Risk in Communities Study (ARIC) 3922 participants were included in the analyses performed by Alonso et al. and followed for over 20 years, 608 AF incident cases were reported [31]. MS measurements of 245 metabolites in serum were performed. Among two secondary bile acids measured (glycocholenate sulfate, glycolithocolate sulfate) only glycocholenate sulfate was associated with AF (per SD, HR 1.12 [95% CI 1.04-1.21]) [29]. Deoxycholic acid, however, was not associated with HF events in the CRIC study [32].
In the Wang et al. study, 136 patients with HF and 51 participants without HF were followed for a mean of 2.3 years and 35 of them had a composite event of death or HF-related re-hospitalization [42]. In the multivariable Cox regression analysis, plasma concentrations of indoxyl sulfate measured by MS were not significantly associated with the composite event.
Significant associations were also reported lately in a large cohort of 5331 participants. During 13.2 years, 4791 deaths were recorded and after adjusting for many potential confounders increased plasma levels of TMAO were associated with an increased risk of all-cause mortality (HR 1.30 [95% CI 1.17-1.44]) [27].
In the Winther et al. study after a median follow-up of 6.8 years and 6.5 years, 106 total deaths and 44 CVD deaths were reported, respectively [18]. No significant associations between one SD increase in TMAO plasma concentrations and any of these outcomes were found in the fully adjusted model. Likewise, in another cohort study conducted by Croyal et al. 1463 patients with T2D were followed for 7.1 years and 538 death cases were recorded. Plasma TMAO was not independently associated with risk of all-cause mortality [22].
Finally, Roncal et al. determined TMAO in plasma using MS in 262 patients with symptomatic PAD who were followed for a mean period of 4 years [38]. During this period, 101 all-cause and 39 CVD mortality cases were reported. A significant association between high TMAO levels and all-cause mortality was found in the unadjusted model, but not in the multivariate-adjusted model. However, higher TMAO levels were associated with CVD-mortality (per one SD increase; HR 1.52 [95% CI 1.27-1.82]) and also when TMAO was treated as a binary variable (> 2.26 µmol/L vs. <2.26 µmol/L; HR 3.36 [95% CI 1.68-6.70]) [38].

Secondary Bile Acids
Yu et al. analyzed 204 serum metabolites in 1887 African Americans from the ARIC Study using MS techniques [36]. During a median follow-up period of 22.5 years 671 deaths were reported. Glycocholate was the only bile acid independently associated with the risk of CVD mortality (HR 1.14 [95% CI 1.07-1.22]) and all-cause mortality (HR 1.12 [95% CI 1.07-1.16]) [36].
Significant results were also obtained for the relationship between deoxycholic acid and all-cause mortality (HR 2.13 [95% CI 1.25-3.64]) in patients with chronic kidney disease [32].
On the other hand, Huang et al. analyzed data from the Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) Study Cohort of 620 men free of CVD, who were followed for 28 years. Of the 406 metabolites quantified in serum using MS techniques no significant associations of the ten secondary bile acids analyzed with CVD or all-cause mortality were observed [30].
No significant associations between three secondary bile acids and all-cause mortality were also reported in an earlier study by Cheng et al. among 2327 participants free of CVD [28].

Tryptophan and Indole Derivatives
Balasubramanian et al. used two datasets created from the Women's Health Initiative study consisting of 943 subjects (discovery set; 417 deaths) and 1355 subjects (replication set; 685 deaths) free of CVD [40]. Of the 470 metabolites measured in plasma with LC-MS/MS the amino acid, tryptophan was inversely associated with all-cause mortality in the discovery (HR 0.82 [95% CI 0.75-0.89]) and replication set (HR 0.87 [95% CI 0.81-0.94]) [40]. No studies examining associations between indole derivatives and mortality were identified in the present review, and more work is needed in this area.

Discussion
This systematic review summarized the evidence of the associations between circulating microbial-related metabolite levels and risk of CVD. Associations of TMAO and subsequent risk of CV outcomes were supported by some [19,20,22] but not all prospective studies [18,21,22,38]. Inconsistent results were also obtained for secondary bile acids in relation to CVD and related outcomes [30,31], and CVD/all-cause mortality [28,30,36]. With regards to BCAAs [39,41], their associations with CV outcomes were robust among the studies, whereas one study reported a negative relationship between tryptophan and mortality [40] and another study revealed an association between indole derivatives and MACE [30].

TMAO
In recent years, several systematic reviews of prospective studies have suggested that TMAO is an independent risk factor for CVD [9][10][11][12][13][14][15][16][17]. Several mechanisms have been described in the literature that could explain these associations. TMAO production from ingestion of animal foods such as red meat may affect cholesterol and lipid metabolism, and endothelial dysfunction as well as platelet activity, which could result in atherosclerosis development [8] the dominant cause of CVD [43]. On the other hand, a recent Mendelian randomization analysis examining the causal direction between TMAO and cardiometabolic diseases suggested that CVD may increase circulating TMAO levels and that observational evidence for CVD may be due to confounding or reverse causality [44].
Our updated systematic review included a total of 11 individual studies that examined the associations of TMAO with CV outcomes. Among them, four outcomes (MACE, major vascular event, CVD mortality, all-cause mortality) were found to be positively associated with TMAO levels in some but not all of the included studies. In three previous meta-analyses of 19, [10], 11 [12] and 17 [11] prospective cohort studies published in 2017, elevated circulating levels of TMAO were associated with higher risk of MACE and allcause mortality with low to moderate and moderate to high heterogeneity, respectively. A more recent systematic review with meta-analysis of three cohort studies also found moderate heterogeneity across the studies examining associations between TMAO and MACE incidence with no evidence of heterogeneity in the association between TMAO and all-cause mortality [13]. Another recent meta-analysis of nine cohort studies revealed that elevated plasma TMAO concentrations were associated with higher risk of MACE evidence of moderate heterogeneity [17]. In our systematic review the discrepancies in the results of the included studies may be due to differences in: disease status, study geographical locations, number of cases, fasting status at blood draw, adjustments for confounders and follow-up length. Disease status might affect the association between TMAO and CV outcomes through affecting microbiota composition which regulate the circulating TMAO levels [45,46]. The role of geographical location in the above associations could be possibly due to the role of diet in regulation of blood TMAO levels [47,48]. Variation in the number of cases among the included studies may have affected statistical power to detect associations. In two studies, blood samples were collected in non-fasting conditions, while in two others fasting status was not indicated which might have affected TMAO concentrations. Finally, uncontrolled confounding factors such as dietary habits, gut microbiota composition and genetic variation may significantly affect the concentrations of TMAO representing an important source of variability in interpretations.

BCAAs
The findings of the present review confirm previous findings of a former systematic review on the strong positive prospective associations between BCAAs and CVD risk [9]. In Tobias et al. study [39], a significant positive association was observed between BCAAs with long-term CVD risk, comparable in magnitude to LDL cholesterol, an established CVD biomarker. Notably, this relationship was prominent among women who developed T2D prior to CVD. Also, in Du et al. [41], increased plasma BCAAs levels were associated with long-term adverse cardiac events in patients with STEMI and AHF and additionally, BCAAs improved the predictive value of NT-proBNP and the GRACE score. These consistent findings reinforce the role of these metabolites as potential biomarkers of biological dysfunction related to CVD.
Potential mechanisms explaining the relationship between elevated BCAAs levels and risk of CVD include activation of the mammalian target of rapamycin (mTOR) signaling, which leads to exacerbated cardiac dysfunction, and remodeling in myocardial infarction [49].

Secondary Bile Acids
To our knowledge, this is the first systematic review of secondary bile acids in relation to CVD risk. Inconsistent results were obtained for these relationships. No significant results were shown for total CVD [30,34], whereas increased levels of glycocholenate sulfate and glycocholate were associated with risk of AF [31], CVD mortality and all-cause mortality [11], respectively. However, the associations of glycocholate, and several other secondary bile acids with mortality were not confirmed in another three studies [27,30,32]. Potential mechanisms underlying the relationship between glycocholenate and AF may include potential arrhythmogenic effects of this bile acid. With regards to glycocholate, is an indirect cholesterol-derived bile acid involved in cholesterol metabolism and is it considered to be a key factor for mortality [50]? Another bile acid, the deoxycholic acid, was also associated with mortality [32]. Elevated levels of this bile acid may contribute to inflammation and immune dysregulation, and vascular calcification and may hasten chronic kidney disease progression.

Tryptophan and Indole Derivatives
In our systematic review tryptophan was inversely associated with all-cause mortality (Women's Health Initiative study) [40]. Tryptophan is an essential amino acid vital for maintaining health and homeostasis [51]. Given that a higher kynurenine/tryptophan ratio reflects an increased indoleamine 2,3-dioxygenase activity and predicts mortality [52] we can speculate that the effect of tryptophan on mortality might depends on its degradation rate to kynurenine and thus higher circulating tryptophan levels may be due to a low rate of conversion.
Of indole derivatives, indoxyl sulfate is a protein-binding molecule that exhibits CV toxicity and in our systematic review was associated with risk of MACE in CKD patients [31]. Some prospective studies suggested that elevated circulating indoxyl sulfate levels may have a significant role in the vascular dysfunction in CKD patients [53]. When dietary tryptophan is metabolized into indole by intestinal bacteria and absorbed, then it is converted to indoxyl sulfate in the liver. In patients with impaired renal function indoxyl sulfate is not removed effectively [54] and accumulates in the blood exerting its deleterious effects on the endothelium by increasing vascular calcification and stiffness [55,56].

SCFA
Altered circulating SCFA profiles may be related to several metabolic conditions. SCFA are produced in the gut by the metabolic activity of the intestinal microbiota as catabolic end-products from the fermentation of undigested dietary components, mainly complex carbohydrates. Although studies have focused on different clinical conditions, evidence on the role of these mediators in CVD is lacking [57].

Strengths and Limitations
To the best of our knowledge, this systematic review provided the most comprehensive evidence on the associations of microbial metabolites with risk of several CV outcomes. Focusing on prospective studies allowed us to limit the influence of reverse causality and selection bias, and all of the included studies were evaluated to be of high quality, as assessed with the "Newcastle-Ottawa Scale".
There are several limitations to be considered. First, the lack of studies on SCFAs limited conclusions about the role of main metabolites produced by intestinal bacteria in CVD and this is an area that needs development. Second, due to the observational nature of the studies included in the review, potential causal mechanisms underlying the above associations cannot be inferred. Third, confounder adjustment differed across the studies, which may have resulted in different associations between metabolites such as TMAO or bile acids and CV outcomes. Fourth, we were unable to conduct a meta-analysis, mainly because of the heterogeneity of the included articles. Differences in metabolite targets, characteristics of the study populations and differential reporting of associations could increase the possibility the results of the meta-analysis not to be useful and meaningful.

Conclusions
In conclusion, our systematic review summarized the evidence on the associations of gut microbiota-derived metabolites with CV outcomes and showed inconsistent results for TMAO and bile acids but robust ones for the relationships between BCAAs and CVD. There are still a small number of longitudinal studies assessing the association of tryptophan and indole derivatives with the risk of CV outcomes. Further studies are needed to investigate whether circulating microbial metabolites could be an intervention target for CVD.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/nu14132654/s1, Table S1: PubMed search strategy, Table S2: Quality assessment of the prospective cohort studies, based on the NEWCASTEL-OTTAWA criteria, Table S3: Quality assessment of the nested case-control studies, based on the NEWCASTEL-OTTAWA criteria.

Data Availability Statement:
No new data were created or analyzed in this study. Data sharing is not applicable to this article.

Conflicts of Interest:
The authors declare no conflict of interest.