Dose-response meta-analysis of plasma TMAO and stroke: validated linear risk threshold at 3.0 μmol/L

Abstract

Background:

Stroke, especially the ischemic type, remains a leading global cause of death and disability, with modifiable risk factors offering prevention opportunities. Trimethylamine N-oxide (TMAO), a gut-derived metabolite, promotes vascular damage and is linked to stroke risk. Although prior studies have explored dose-response relationships, clinically actionable thresholds remain undefined, limiting translational applications. This study aims to advance the field by quantifying a continuous dose-response relationship and determining a specific risk threshold, which is currently lacking, to inform preventive strategies.

Methods:

This PRISMA-compliant meta-analysis included 11 observational studies (n = 7,556) and encompassed two components: an overall meta-analysis of 10 studies to compare admission TMAO levels, and a dose-response meta-analysis that was specifically applied to the subset of 4 studies with sufficient data across multiple exposure categories. We pooled standardized mean differences (SMD) for admission TMAO levels and modeled dose-response curves using restricted cubic splines (knots at 2.37/3.45/5.95 μmol/L). Heterogeneity was quantified using the I²-statistic, sensitivity was assessed using alternative statistical models and dose scaling approaches, and publication bias was evaluated with Egger's test and the trim-and-fill method.

Results:

Stroke patients showed significantly higher TMAO vs. controls (SMD = 0.55, 95% CI: 0.35, 0.74; P < 0.00001). Linear dose-response relationship: Each 1 μmol/L TMAO increase raised stroke risk by 8.9% (OR = 1.089, 95% CI: 1.023–1.158; P = 0.007). Risk threshold: TMAO > 3.0 μmol/L significantly increases the risk (OR > 1) and warrants preventive intervention. Cumulative risk escalated: 0 → 5 μmol/L: 53% risk increase (OR = 1.53); 0 → 20 μmol/L: 448% risk increase (OR = 5.48); robustness confirmed by sensitivity analysis (I² = 35.9%; Cochran Q, P = 0.154).

Conclusion:

TMAO exhibits a linear, dose-dependent association with stroke risk, with ≥ 3.0 μmol/L serving as a critical threshold for clinical intervention.

1 Introduction

Stroke is categorized into ischemic and hemorrhagic subtypes, with approximately 87% of cases being ischemic (1). Ischemic stroke (IS) accounts for 80% of all strokes and is one of the most fatal diseases globally (2). The fundamental pathological cause of IS is intravascular thrombosis, leading to cerebral tissue necrosis and focal neuronal deficits (3). Globally, IS caused 188.3 million disability-adjusted life years (DALYs) in 2021, positioning it as the second leading cause of death (4). In the US, IS accounts for approximately 87% of all strokes, representing the fifth leading cause of death and a primary contributor to severe long-term disability (5). In China, incidence surged from 760,000 cases (1990) to 2.77 million cases (2021), the age-standardized rate of DALYs decreased by 0.50%, but the absolute number rose (from 9.92 million to 23.43 million) (6, 7).

Trimethylamine N-oxide (TMAO) is a gut microbiota-derived metabolite. Dietary nutrients rich in phosphatidylcholine and choline—such as red meat, eggs, and to a lesser extent, dairy products like milk—are metabolized by gut microbes into trimethylamine (TMA). TMA enters the portal circulation and is oxidized to TMAO in the liver, primarily by flavin-containing monooxygenases (FMO), especially FMO3 (8, 9). TMAO is eliminated from the body primarily via renal excretion and secondarily through reduction back to TMA by intestinal microbes (10). Studies indicate that renal function is a critical determinant of TMAO accumulation, even mild renal impairment (eGFR < 66 ml/min/1.73 m²) can lead to elevated levels of TMAO (11, 12). TMAO exhibits strong associations with cardiovascular and cerebrovascular pathologies. Elevated TMAO levels are associated with reduced expression of tight junction proteins (claudin-5, ZO-1) in experimental models, which may contribute to cerebral small vessel damage (10). Moreover, TMAO promotes atherosclerosis by upregulating scavenger receptors (CD36/SR-A1) on macrophages, thereby accelerating foam cell formation (10, 13, 14). It also positively correlates with pro-inflammatory monocytes (CD14++CD16+; r = 0.70), potentially exacerbating thrombosis and vascular inflammation through monocyte activation (15, 16).

Based on the aforementioned pathophysiological mechanisms, several meta-analyses have supported the TMAO-stroke association (17–19), but they yield conflicting conclusions regarding the dose-response shape. Specifically, the meta-analyses by Farhangi et al. (18) identified a non-linear relationship, whereas the meta-analysis by Chen et al. (17) and Hong et al. (19) found a linear one. Building upon prior meta-analyses that identified an association, this study aims to advance the field by a dose-response meta-analysis (DRMA) to determine a specific risk threshold, which is currently lacking, to inform preventive strategies.

2 Materials and methods

2.1 Search strategy

This DRMA was conducted and reported in accordance with the PRISMA guidelines (20). Literature searches encompassed records from database inception until 1 July 2025, with language restrictions to English or Chinese. The study details have been registered with PROSPERO under number CRD420251151002.

We searched the following databases: PubMed, Web of Science (WOS), Embase, Scopus, Cochrane Library, Sinomed, China National Knowledge Infrastructure (CNKI), WanFang Data, and VIP (China Science and Technology Journal Database). Clinical trial registries: ClinicalTrials.gov and the Chinese Clinical Trial Registry (ChiCTR) to identify ongoing or unpublished trials.

Search terms included TMAO, trimethylamine N-oxide, stroke, cerebrovascular accident, CVA, observational study, cohort Study, cross-sectional Study, case-control study, etc. The detailed search strategy is provided in the Supplementary Materials, and a flow diagram is presented in Figure 1.

Figure 1

2.2 Inclusion and exclusion criteria

Inclusion criteria: (a) Observational studies (cohort, case-control, or cross-sectional designs); (b) patients with a confirmed stroke diagnosis based on neuroimaging (e.g., MRI or CT), and subsequently classified by etiology using standardized systems such as the TOAST criteria; (c) measurement of blood TMAO levels; (d) provision of accessible TMAO level statistics [e.g., mean/standard deviation (SD) or median/interquartile range (IQR)].

Exclusion criteria: (a) Randomized controlled trials involving TMAO interventions (e.g., drug or dietary interventions) or stroke treatment, as well as dissertations, reviews, commentaries, conference abstracts, case reports, correspondence articles, and animal or cell experimental studies; (b) studies without a clear diagnosis of “stroke” or using unrecognized diagnostic criteria; (c) studies with unreliable results due to unadjusted or incompletely adjusted confounding variables; (d) studies with inaccessible full texts, incomplete data reporting, or duplicate publications; (e) studies with a sample size of < 50 per group.

2.3 Study selection and data extraction

Two authors (JZ and LL) independently searched databases and screened titles, abstracts, and full texts to identify eligible articles. Disagreements were resolved through discussion with a third author (RL). Data extraction by two authors (JZ and TY) followed a predefined protocol, capturing information on title, author, country, publication year, group characteristics, outcome measures, results, and conclusions. Disagreements during extraction were resolved by consulting the third author (RL).

2.4 Risk of bias assessment

Two reviewers (TY and LL) independently assessed the methodological quality of the included observational studies using the Newcastle–Ottawa Scale (NOS) (21). This tool evaluates studies based on three domains: selection of the study groups (maximum 4 stars), comparability of the groups (maximum 2 stars), and ascertainment of either the outcome or exposure (maximum 3 stars), yielding a total maximum score of 9. A higher score indicates higher study quality, with a score of ≥7 typically indicating a high-quality study. Any disagreements between the reviewers during the assessment were resolved through discussion or adjudication by a third reviewer (RL).

2.5 Data analysis

In line with the primary objective of establishing a dose-response relationship, the data analysis proceeded in two stages. First, a standard meta-analysis was performed using Review Manager 5.4 software. Data were presented as means () and SD with 95% CIs. A fixed-effects model (FEM) was used if I² < 50% (no significant heterogeneity); otherwise, a random-effects model (REM) was employed. For studies (22–28) reporting medians and IQR, data were converted to and SD using formulas of Wan et al. (29) and Luo et al. (30): , (IQR = q₃−q₁).

Second, and central to this aim, a DRMA was performed on 4 studies (22, 23, 26, 31) that provided sufficient data on multiple exposure categories, enabling the examination of concentration-response relationships. This analysis was performed using R 4.5.1 with a two-stage random-effects framework. Within-study non-independence across dose groups was addressed using Greenland–Longnecker covariance correction (32). Studies-specific dose-response curves were fitted using weighted least squares. A restricted cubic spline (RCS) model with three knots (positioned at the 10th, 50th, and 90th percentiles of the dose distribution: 2.37, 3.45, 5.95 μmol/L) modeled non-linear relationships. In the between-study stage, maximum likelihood (ML) was used for fixed effects, and restricted maximum likelihood (REML) was used to estimate the random-effects covariance structure. Especially, a key finding of our analysis was the determination of a TMAO risk threshold, defined as the lowest dose at which the lower bound of the 95% CI for the pooled OR first exceeded 1.0. Specifically, the algorithm automatically identified this threshold by scanning dose points in ascending order and selecting the first dose that met this criterion. Heterogeneity was quantified using Cochran's Q test (P > 0.10 indicating low heterogeneity) and the I²-statistic (>50% indicating substantial heterogeneity). Traditional leave-one-out sensitivity analysis was precluded, as the study by Nie et al. (23) contributed only two non-referent observations, rendering model fitting unstable upon exclusion of any study. We therefore assessed robustness using alternative strategies, including different statistical methods (fixed-effect and maximum-likelihood models) and dose transformations (log and standardization). Publication bias was assessed visually via funnel plots and quantitatively using Egger's linear regression test. Trim-and-fill analysis was used for bias correction.

3 Results

3.1 Characteristics of included studies

11 studies (22–28, 31, 33–35) (total n = 7,556; 3,878 stroke patients, 3,678 controls) published between 2017 and 2023 were included. 9 studies (22–24, 26–28, 33–35) originated from China, 1 from Germany (25), and 1 from the Spain (31). TMAO was primarily measured via liquid chromatography-tandem mass spectrometry (LC-MS/MS), with 1 study (24) using ultra-high-performance LC-MS/MS (UHPLC-MS/MS). Stroke types, all confirmed by neuroimaging (MRI/CT), included ischemic stroke (24–28, 33–35) and mixed types (22, 23, 31). The TOAST criteria were used for etiological classification of the ischemic strokes. The characteristics of the included studies are presented in Table 1.

3.2 Risk of bias

The quality of the 11 included observational studies was evaluated using the NOS. The assessment showed that the NOS scores of all studies ranged from 6 to 8 out of a maximum of 9, indicating satisfactory overall methodological quality. Specifically, 9 studies (22–24, 26, 27, 31, 33–35) were rated as high quality (NOS score ≥ 7), while the remaining 2 studies (25, 28) had a score of 6, indicating moderate quality. Most studies scored highly in the “Selection” and “Outcome” (or “Exposure”) domains, reflecting rigorous definitions and ascertainment of cases and controls, as well as robust assessment methods. Points were primarily deducted in the “Comparability” domain, largely because some studies (25, 28, 33) did not adequately report on or control for key confounding variables (e.g., age, renal function). The detailed results of the quality assessment are presented in Supplementary Table 1.

3.3 Admission TMAO levels

10 studies (22–28, 33–35) reported blood TMAO levels at admission. Significant heterogeneity existed (I² = 93%, likely stemming from different TMAO measurements and admission conditions), warranting a REM. Standardized Mean Difference (SMD) was used as the summary effect measure to better accommodate potential differences in measurement scales of raw data and uncertainties introduced by conversion [of these 10 studies, 6 (23–28) presented the admission TMAO levels by medians and quartiles, which were converted to the mean and SD by formulas of Wan et al. and Luo et al.]. Meta-analysis revealed significantly higher TMAO levels in stroke patients compared to controls (SMD = 0.55, 95% CI: 0.35, 0.74; Z = 5.42, P < 0.00001). Heterogeneity (I² > 50%) likely stems from methodological variations in TMAO measurement and differences in the severity of patients' conditions at admission (Figure 2).

Figure 2

3.4 Dose-response meta-analysis

3.4.1 Primary results

4 studies (22, 23, 26, 31) were included in the DRMA. A significant dose-response relationship existed between TMAO and stroke risk (χ² = 13.6170 (df = 2), P = 0.0011). Linear dose-effect: Each 1 μmol/L increase in TMAO was associated with a significant 0.0851 increase in the log-odds ratio (logOR) for stroke (95% CI: 0.0232, 0.1469; P = 0.0070), corresponding to an OR of 1.089 (95% CI: 1.023, 1.158). This indicates an 8.9% increased stroke risk per 1 μmol/L TMAO increment. Non-linearity was non-significant (P = 0.9479, 95% CI: −0.2012, 0.1883), supporting a linear trend. Moderate heterogeneity was observed (I² = 35.9%; Cochran Q: Q = 9.3647, df = 6, P = 0.1541) (Detailed data are presented in Table 2). The RCS plot for the dose-response relationship is presented in Figure 3. Following the establishment of a significant dose-response relationship (model chi-square test P = 0.001), a key finding of our analysis was the identification of a TMAO risk threshold at 3.0 μmol/L. In addition, we calculated the cumulative risk increase and plot a radar chart (Figure 4) to visually demonstrate this risk accumulation.

Figure 3

Figure 4

3.4.2 Sensitivity analysis

Sensitivity analyses confirmed the robustness of the primary finding. As shown in Figure 5, all alternative approaches yielded a consistently positive association. The results from the FEM (β = 0.074) and maximum likelihood method (β = 0.076) were highly consistent with the primary analysis (β = 0.085). While point estimates from log-dose transformation (β = 0.222) and dose standardization (β = 0.216) were higher—reflecting the change in units of measurement—the direction of association remained unchanged. In summary, all pre-specified sensitivity analyses yielded results that corroborated the primary analysis, underscoring the robustness of the association between elevated blood TMAO levels and an increased risk of stroke.

Figure 5

3.4.3 Publication bias

Publication bias was assessed using a funnel plot (Figure 6) and Egger's test. Funnel plot visual inspection suggested symmetry. Egger's test showed borderline significance (Detailed data are presented in Table 2), potentially indicating limited power due to the small number of studies (n = 4). Further verification using the trim-and-fill (Figure 7) analysis indicated no missing studies (0 imputed, SE = 1.5649), yielding a consistent corrected slope β = 0.0931 (95% CI: 0.019, 0.167, P = 0.0134) (Detailed data are presented in Table 2).

Figure 6

Figure 7

4 Discussion

This study, utilizing a DRMA, has for the first time established and quantified a significant linear dose-response relationship between plasma TMAO levels and the risk of stroke. More importantly, we identified a critical clinical risk threshold, beyond which the stroke risk begins to demonstrate a statistically significant increase.

This finding holds substantial clinical translational value: it provides, for the first time, a quantifiable TMAO-based intervention threshold for stroke prevention. This enables clinicians to utilize TMAO as an objective biomarker for risk stratification in high-risk populations, facilitates the development of personalized prevention strategies, and allows for the dynamic assessment of the efficacy of interventions (such as dietary modifications or pharmacological treatments) during patient management. Wu et al. (36) reported a comparable threshold (4.29 μmol/L) in post-carotid stenting patients. While higher than our 3.0 μmol/L, both results confirm TMAO's promise as a risk marker for intervention. Additionally, multiple independent studies (17, 19) consistently report that each unit increase in TMAO significantly elevates stroke risk, consistent with our linear relationship. Among which, Chen and Guo (17), in their “Dose-Response Analysis” section, did test the possibility of a non-linear association. However, the statistical test result (P for non-linearity = 0.0738) was not significant. But a study (18) identified a non-linear relationship between TMAO and stroke risk, which may be attributed to the high heterogeneity of included studies (I² = 99.7%, encompassing diverse populations, stroke types, and geographical regions) and the wide range of TMAO concentrations (0–20 μmol/L). In contrast, this study is the first to validate a specific linear risk threshold through a rigorous DRMA, which provides a clear intervention target for clinical practice, enabling physicians to stratify stroke risk and make preventive decisions based on TMAO levels, while other studies only reported risk increments or non-linear trends, lacking directly applicable thresholds. Secondly, the use of a two-stage random-effects model, Greenland–Longnecker covariance correction, and restricted cubic spline analysis ensured the robustness of the dose-response relationship. This robustness was further supported by sensitivity analyses. An exploratory assessment for publication bias was conducted (e.g., Egger's test), the interpretation of which is limited by the small number of studies. Notably, the pathophysiological role of TMAO extends beyond this. Hou et al. (37) found that elevated TMAO at admission independently predicts early neurological deterioration after stroke. Combined evidence positions TMAO as a persistent biomarker throughout stroke: lower pre-stroke thresholds (e.g., 3.0 μmol/L) may be applicable for primary prevention and risk stratification, while elevated levels in the acute phase may predict a worse clinical course.

In sensitivity analyses employing different statistical methods and dose-scaling strategies, the positive association between TMAO and stroke risk remained consistent, indicating that the primary finding is robust and not overly dependent on specific analytical assumptions. An apparent inconsistency was observed in the study by Guasch-Ferré et al. (31), which reported a decrease from an OR of 1.19 in Q3 to 0.98 in Q4. However, the wide and overlapping confidence intervals (Q3: 0.65–2.17; Q4: 0.53–1.84) suggest that this difference is not statistically significant and likely reflects random variation. Furthermore, the unique characteristics of the Mediterranean study population may also have contributed to this result. Their dietary patterns (rich in olive oil, nuts, and fish) may provide direct TMAO precursors (non-microbiota-dependent sources of TMAO) (38, 39). As noted by Caradonna et al. (8), diet custom can alter TMAO metabolism, explaining the heterogeneity in Mediterranean-diet population. Moreover, Hu et al. (40) also reported a non-significant association between TMAO levels and stroke incidence in the European subgroup. Publication bias was assessed. The funnel plot exhibited symmetry, and Egger's test was not significant, suggesting a low risk of bias. Moreover, the trim-and-fill correction did not impute any missing studies, and the effect size remained consistent after this procedure, further supporting the robustness of the results. Moderate heterogeneity, possibly due to population or methodological variations, did not affect the core conclusion. The positive dose-response relationship between TMAO and stroke risk is robust.

The strengths of this study include the establishment of a quantified linear dose-response relationship between TMAO and stroke risk, with 3.0 μmol/L identified as a clinical risk threshold. This finding was supported by robust statistical modeling utilizing a two-stage random-effects model with Greenland–Longnecker correction and restricted cubic spline testing. The analysis further demonstrated a clear risk stratification, revealing an accelerating risk gradient. Methodological rigor was ensured by excluding studies with sample sizes under 50 or incomplete adjustment for confounders, and potential publication bias was minimized through the application of trim-and-fill analysis.

However, this study is subject to several limitations. First, there is a notable regional bias in the included studies, with 9 out of 11 studies originating from China. Factors such as specific dietary patterns (e.g., Asian vs. Mediterranean diets) and gut microbiota composition may restrict the global generalizability of our findings. Second, there was insufficient control for potential confounders; for instance, acute-phase TMAO levels could be influenced by renal function or medication use, yet the included studies lacked standardized adjustment for these variables. Third, the dose-response analysis included only four studies, resulting in limited statistical power. Fourth, heterogeneity also stems from methodological differences in TMAO measurement, including the use of plasma vs. serum samples and varying timing of blood collection in the acute stroke phase. Finally, the literature search was restricted to Chinese and English databases, potentially introducing language bias.

Furthermore, this study could not explore diet-specific risk factors. Critically, the original studies included in our meta-analysis did not provide data on meat intake or other dietary components. Therefore, we could not directly establish the association between meat consumption and stroke risk within this analysis. Nonetheless, given that TMAO is primarily derived from dietary phosphatidylcholine and L-carnitine (abundant in red meat and processed meat), our results indirectly support the potential strategy of changing diet (e.g., reducing meat intake) to lower TMAO levels and thereby prevent stroke.

5 Conclusion

In summary, our analysis indicates a linear dose-response relationship between TMAO and stroke risk, identifying 3.0 μmol/L as a potential intervention threshold. Given the limitations such as regional bias and methodological variations, this finding requires further validation. Future research should prioritize standardized, large-scale multi-center studies to confirm these results and assess the clinical utility of TMAO monitoring in stroke prevention.

References

• 1.

  Tsao CW Aday AW Almarzooq ZI Alonso A Beaton AZ Bittencourt MS et al . Heart disease and stroke statistics-−2022 update: a report from the American Heart Association. Circulation. (2022) 145:e153–639. doi: 10.1161/CIR.0000000000001052

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2.

  Amarenco P Bogousslavsky J Caplan LR Donnan GA Hennerici MG . Classification of stroke subtypes. Cerebrovasc Dis. (2009) 27:493–501. doi: 10.1159/000210432

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3.

  Zhao Y Zhang X Chen X Wei Y . Neuronal injuries in cerebral infarction and ischemic stroke: from mechanisms to treatment (review). Int J Mol Med. (2021) 49:15. doi: 10.3892/ijmm.2021.5070

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4.

  Ferrari AJ Santomauro DF Aali A Abate YH Abbafati C Abbastabar H et al . Global incidence, prevalence, years lived with disability (YLDs), disability-adjusted life-years (DALYs), and healthy life expectancy (HALE) for 371 diseases and injuries in 204 countries and territories and 811 subnational locations, 1990–2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet. (2024) 403:2133–61. doi: 10.1016/S0140-6736(24)00757-8

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5.

  Petty K Lemkuil BP Gierl B . Acute ischemic stroke. Anesthesiol Clin. (2021) 39:113–25. doi: 10.1016/j.anclin.2020.11.002

  • CrossRef
  • Google Scholar

• 6.

  Hou S Zhang Y Xia Y Liu Y Deng X Wang W et al . Global, regional, and national epidemiology of ischemic stroke from 1990 to 2021. Eur J Neurol. (2024) 31:e16481. doi: 10.1111/ene.16481

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7.

  Vollset SE Ababneh HS Abate YH Abbafati C Abbasgholizadeh R Abbasian M et al . Burden of disease scenarios for 204 countries and territories, 2022–2050: a forecasting analysis for the Global Burden of Disease Study 2021. Lancet. (2024) 403:2204–56. doi: 10.1016/S0140-6736(24)00685-8

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 8.

  Caradonna E Abate F Schiano E Paparella F Ferrara F Vanoli E et al . Trimethylamine-N-oxide (TMAO) as a rising-star metabolite: implications for human health. Metabolites. (2025) 15:220. doi: 10.3390/metabo15040220

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9.

  Zhu Y Sun Y Hu J Pan Z . Insight into the mechanism of exercise preconditioning in ischemic stroke. Front Pharmacol. (2022) 13:866360. doi: 10.3389/fphar.2022.866360

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10.

  Paulina G Kaluzna-Czaplinska J . Trimethylamine N-oxide (TMAO) in human health. EXCLI J. (2021) 20:301–19. doi: 10.17179/excli2020-3239

  • CrossRef
  • Google Scholar

• 11.

  Spence JD . Trimethylamine N-oxide: not just red meat—egg yolk and renal function are also important. Eur Heart J. (2019) 40:3498. doi: 10.1093/eurheartj/ehz615

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12.

  Wang M Li XS Wang Z de Oliveira Otto MC Lemaitre RN Fretts A et al . Trimethylamine N-oxide is associated with long-term mortality risk: the multi-ethnic study of atherosclerosis. Eur Heart J. (2023) 44:1608–18. doi: 10.1093/eurheartj/ehad089

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13.

  Li Z He X Fang Q Yin X . Gut microbe-generated metabolite trimethylamine-N-oxide and ischemic stroke. Biomolecules. (2024) 14:1463. doi: 10.3390/biom14111463

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14.

  Nam HS . Gut microbiota and ischemic stroke: the role of trimethylamine N-oxide. J Stroke. (2019) 21:151–9. doi: 10.5853/jos.2019.00472

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15.

  Haghikia A Li XS Liman TG Bledau N Schmidt D Zimmermann F et al . Gut microbiota–dependent trimethylamine N-oxide predicts risk of cardiovascular events in patients with stroke and is related to proinflammatory monocytes. Arterioscler Thromb Vasc Biol. (2018) 38:2225–35. doi: 10.1161/ATVBAHA.118.311023

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16.

  Zhen J Zhou Z He M Han H-X Lv E-H Wen P-B et al . The gut microbial metabolite trimethylamine N-oxide and cardiovascular diseases. Front Endocrinol. (2023) 14:1085041. doi: 10.3389/fendo.2023.1085041

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17.

  Chen P Guo Z . Plasmatic trimethylamine N-oxide and its relation to stroke: a systematic review and dose-response meta-analysis. Medicine. (2022) 101:e29512. doi: 10.1097/MD.0000000000029512

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18.

  Farhangi MA Vajdi M Asghari-Jafarabadi M . Gut microbiota-associated metabolite trimethylamine N-oxide and the risk of stroke: a systematic review and dose–response meta-analysis. Nutr J. (2020) 19:76. doi: 10.1186/s12937-020-00592-2

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19.

  Hong Y Sun Z Liu N Yang K Li Y Xu Q et al . The relationship between trimethylamine-N-oxide and the risk of acute ischemic stroke: a dose–response meta-analysis. PLoS ONE. (2023) 18:e0293275. doi: 10.1371/journal.pone.0293275

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20.

  Page MJ McKenzie JE Bossuyt PM Boutron I Hoffmann TC Mulrow CD et al . The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Syst Rev. (2021) 10:89. doi: 10.1186/s13643-021-01626-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21.

  Wells GA Shea B O'Connell D Peterson J Welch V Losos M et al . The Newcastle-Ottawa Scale (NOS) for Assessing the Quality of Nonrandomised Studies in Meta-Analyses (2024). Available online at: https://www.ohri.ca/programs/clinical_epidemiology/oxford.asp (Cited September 12, 2025).

  • Google Scholar

• 22.

  Liu D Gu S Zhou Z Ma Z Zuo H . Associations of plasma TMAO and its precursors with stroke risk in the general population: a nested case-control study. J Intern Med. (2022) 293:110–20. doi: 10.1111/joim.13572

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23.

  Nie J Xie L Zhao B-x Li Y Qiu B Zhu F et al . Serum trimethylamine N-oxide concentration Is positively associated with first stroke in hypertensive patients. Stroke. (2018) 49:2021–8. doi: 10.1161/STROKEAHA.118.021997

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24.

  Rexidamu M Li H Jin H Huang J . Serum levels of trimethylamine-N-oxide in patients with ischemic stroke. Biosci Rep. (2019) 39:BSR20190515. doi: 10.1042/BSR20190515

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 25.

  Schneider C Okun JG Schwarz KV Hauke J Zorn M Nürnberg C et al . Trimethylamine-N-oxide is elevated in the acute phase after ischaemic stroke and decreases within the first days. Eur J Neurol. (2020) 27:1596–603. doi: 10.1111/ene.14253

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26.

  Sun T Zhang Y Yin J Peng X Zhou L Huang S et al . Association of gut microbiota-dependent metabolite trimethylamine N-oxide with first ischemic stroke. J Atheroscler Thromb. (2021) 28:320–8. doi: 10.5551/jat.55962

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27.

  Wu C Xue F Lian Y Zhang J Wu D Xie N et al . Relationship between elevated plasma trimethylamine N-oxide levels and increased stroke injury. Neurology. (2020) 94:e667–77. doi: 10.1212/WNL.0000000000008862

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28.

  Zhang J Wang L Cai J Lei A Liu C Lin R et al . Gut microbial metabolite TMAO portends prognosis in acute ischemic stroke. J Neuroimmunol. (2021) 354:577526. doi: 10.1016/j.jneuroim.2021.577526

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29.

  Xiang Wan Wenqian Wang Jiming Liu Tong T . Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med Res Methodol. (2014) 14:135. doi: 10.1186/1471-2288-14-135

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30.

  Luo D Wan X Liu J Tong T . Optimally estimating the sample mean from the sample size, median, mid-range, and/or mid-quartile range. Stat Methods Med Res. (2016) 27:1785–805. doi: 10.1177/0962280216669183

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31.

  Guasch-Ferré M Hu FB Ruiz-Canela M Bulló M Toledo E Wang DD et al . Plasma metabolites from choline pathway and risk of cardiovascular disease in the predimed (prevention with Mediterranean diet) study. J Am Heart Assoc. (2017) 6:e006524. doi: 10.1161/JAHA.117.006524

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32.

  Greenland S Longnecker MP . Methods for trend estimation from summarized dose-response data, with applications to meta-analysis. Am J Epidemiol. (1992) 135:1301–9. doi: 10.1093/oxfordjournals.aje.a116237

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 33.

  Chen Y-Y Ye Z-S Xia N-G Xu Y . TMAO as a novel predictor of major adverse vascular events and recurrence in patients with large artery atherosclerotic ischemic stroke. Clin Appl Thromb Hemost. (2022) 28:1–7. doi: 10.1177/10760296221090503

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34.

  Xu D Zhao W Song J Yin L Wang K Wei L et al . The relationship of large-artery atherothrombotic stroke with plasma trimethylamine N-oxide level and blood lipid-related indices: a cross-sectional comparative study. BioMed Res Int. (2021) 2021:5549796. doi: 10.1155/2021/5549796

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35.

  Zhu C Li G Lv Z Li J Wang X Kang J et al . Association of plasma trimethylamine-N-oxide levels with post-stroke cognitive impairment: a 1-year longitudinal study. Neurol Sci. (2019) 41:57–63. doi: 10.1007/s10072-019-04040-w

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36.

  Wu C Li C Zhao W Xie N Yan F Lian Y et al . Elevated trimethylamine N-oxide related to ischemic brain lesions after carotid artery stenting. Neurology. (2018) 90:e1283–e90. doi: 10.1212/WNL.0000000000005298

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37.

  Hou L Zhang Y Zheng D Shi H Zou C Zhang H et al . Increasing trimethylamine N-oxide levels as a predictor of early neurological deterioration in patients with acute ischemic stroke. Neurol Res. (2020) 42:153–8. doi: 10.1080/01616412.2019.1710416

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38.

  Croci S D'Apolito LI Gasperi V Catani MV Savini I . Dietary strategies for management of metabolic syndrome: role of gut microbiota metabolites. Nutrients. (2021) 13:1389. doi: 10.3390/nu13051389

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39.

  Malinowska AM Szwengiel A Chmurzynska A . Dietary, anthropometric, and biochemical factors influencing plasma choline, carnitine, trimethylamine, and trimethylamine-N-oxide concentrations. Int J Food Sci Nutr. (2016) 68:488–95. doi: 10.1080/09637486.2016.1256379

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40.

  Hu X Ren H Cao Y . The association between trimethylamine N-oxide levels and ischemic stroke occurrence: a meta-analysis and Mendelian randomization study. BMC Neurol. (2023) 23:413. doi: 10.1186/s12883-023-03458-2

  • Pubmed Abstract
  • CrossRef
  • Google Scholar