Abstract
Recent studies show a mechanistic link between intestinal microbial metabolism of dietary phosphatidylcholine and coronary artery disease pathogenesis. Concentrations of a proatherogenic gut microbe-generated metabolite, trimethylamine N-oxide (TMAO), predict increased incident cardiovascular disease risks in multiple cohorts. TMAO concentrations are increased in patients with type 2 diabetes mellitus (T2DM), but their prognostic value and relation to glycemic control are unclear.
We examined the relationship between fasting TMAO and 2 of its nutrient precursors, choline and betaine, vs 3-year major adverse cardiac events and 5-year mortality in 1216 stable patients with T2DM who underwent elective diagnostic coronary angiography.
TMAO [4.4 μmol/L (interquartile range 2.8–7.7 μmol/L) vs 3.6 (2.3–5.7 μmol/L); P < 0.001] and choline concentrations were higher in individuals with T2DM vs healthy controls. Within T2DM patients, higher plasma TMAO was associated with a significant 3.0-fold increased 3-year major adverse cardiac event risk (P < 0.001) and a 3.6-fold increased 5-year mortality risk (P < 0.001). Following adjustments for traditional risk factors and high-sensitivity C-reactive protein, glycohemoglobin, and estimated glomerular filtration rate, increased TMAO concentrations remained predictive of both major adverse cardiac events and mortality risks in T2DM patients [e.g., quartiles 4 vs 1, hazard ratio 2.05 (95% CI, 1.31–3.20), P < 0.001; and 2.07 (95% CI, 1.37–3.14), P < 0.001, respectively].
Fasting plasma concentrations of the proatherogenic gut microbe-generated metabolite TMAO are higher in diabetic patients and portend higher major adverse cardiac events and mortality risks independent of traditional risk factors, renal function, and relationship to glycemic control.
Choline is an essential nutrient that is both synthesized endogenously and obtained from dietary sources, especially phosphatidylcholine (lecithin), in a variety of animal and plant products (1). Animal model studies reveal adequate levels of choline are required for appropriate formation of phospholipid membranes and the neurotransmitter acetylcholine, with choline deficiency, producing neurological impairment (2). The direct oxidation product of choline is betaine, which is a methyl donor in homocysteine remethylation, and also plays an important role in maintaining cellular volume and stability (3). Together, choline and betaine have been shown to provide hepatoprotection and improve insulin resistance in nonalcoholic fatty liver mouse models (4).
Recent studies indicate a pathophysiological contribution of gut microbiota to cardiometabolic diseases with mechanistic links to gut microbial choline metabolism. Specifically, the gut microbiome has been shown to serve as an important modulator of host energy balance and metabolism in type 2 diabetes mellitus (T2DM)5 (5–7). Our group discovered plasma concentrations of 3 metabolites of dietary phosphatidylcholine—choline, betaine, and trimethylamine N-oxide (TMAO)—are associated with atherosclerotic cardiovascular risks in patients, and mechanistically linked to atherosclerosis development in animal models (8). TMAO formation in both animal models and humans requires gut microbes (8–11), and proceeds via gut microbe metabolism of either choline, betaine, or other trimethylamine (TMA) nutrients to initially form TMA, which is delivered to the liver via the portal circulation where it is rapidly converted by host hepatic flavin monooxygenase 3 into TMAO (8, 12–14). TMAO itself promotes proatherosclerotic biological activities (8, 11, 15), and studies in a large independent cohort revealed the adverse prognostic value of increased plasma choline and betaine concentrations is driven by their nutrient precursor→product relationship with TMAO (14). Interestingly, recent animal model studies using the liver-specific insulin receptor–null mouse in the LDL receptor–null background, an insulin resistance mouse model of atherosclerosis, showed that suppression of hepatic flavin monooxygenase 3 and TMAO levels inhibited both hyperglycemia and atherosclerosis development (16). Moreover, in recent studies, development of an inhibitor of gut microbial formation of TMA, the precursor of TMAO, was shown to inhibit diet-induced atherosclerosis. Previous studies have suggested that patients with T2DM may have increased TMAO concentrations compared to patients without T2DM (17); however, whether or not this relationship was confounded by glycemic control or renal function is unclear. We hypothesize that fasting plasma concentrations of the proatherogenic gut microbe-generated metabolite TMAO portend higher major adverse cardiac events and mortality risks independent of traditional risk factors, renal function, and relationship to glycemic control.
Methods
We prospectively enrolled 2 cohorts for this study. The first comprised sequential stable patients at Cleveland Clinic undergoing elective, nonurgent, coronary angiographic evaluation between 2001 and 2007. We excluded those who had recently experienced acute coronary syndrome within the preceding 30 days (cardiac troponin I ≥ 0.03 ng/mL). In this analysis, we analyzed glucose and glycohemoglobin (Hb A1c), along with available laboratory results and medication use in the electronic medical record, and identified patients with a history of T2DM [defined as documented Hb A1c ≥6.5%, fasting glucose ≥126 mg/dL (6.99 mol/L), oral glucose tolerance test with 2-h postload glucose ≥200 mg/dL (11.1 mmol/L), and/or a history of T2DM with appropriate glucose-lowering drug use]. The second cohort examined was an independent set of 300 prospectively recruited, apparently healthy individuals ≥21 years old without known T2DM or cardiac diseases from health screens in the Cleveland area, solely for the purposes of providing reference interval values of these metabolites. Both studies were approved by the Cleveland Clinic Institutional Review Board.
After informed consent, fasting plasma blood samples were collected using EDTA tubes, and immediately processed and frozen at −80 °C until analysis. Quantification of fasting plasma TMAO concentrations was performed utilizing stable isotope dilution liquid chromatography with online LC-MS/MS on an AB Sciex API 5500 triple quadrupole mass spectrometer (Applied Biosystems) using d9(trimethyl)TMAO (d9-TMAO) as internal standards and detection of precursor product ion transitions at m/z 76→58, as previously described (18). The assay described shows good inter- and intraday reproducibility (all CVs < 7%), accuracy (>98.5% across low, mid and high values), and with good stability with freeze–thaw cycles (≥5, intercycle CV% < 9) (18). Plasma choline and betaine can also be measured simultaneously by the use of this method.
High-sensitivity C-reactive protein (hsCRP), fasting lipid panel, Hb A1c, glucose, insulin, and serum creatinine were measured using the Roche Cobas 4000 analyzer platform (Roche Diagnostics). Estimated glomerular filtration rate (eGFR) was calculated by the Modified Diet in Renal Diseases equation (19). Adverse events were tracked prospectively with direct contact for 3 years for major adverse cardiac events [death, nonfatal myocardial infarction (MI), and nonfatal stroke] using a combination of postcard and scripted telephone follow-up, and manual chart review, as well as all-cause mortality for 5 years by electronic chart review and Social Security Death Index up to 2011 and confirmed by telephone interviews, official hospital records, or death certificates.
Continuous variables are summarized as mean (SD) if normally distributed and median [interquartile range (IQR)] if non-normally distributed. The Student t-test or Wilcoxon signed rank test for continuous variables and χ2 test for categorical variables were employed to examine between group differences. Spearman correlation was used to examine the associations between choline, betaine, TMAO, and other laboratory measurements. Kaplan–Meier survival plots were calculated from baseline to time of adverse event and compared using the log-rank test. Cox proportional hazards analysis were used to determine hazard ratio (HR) and 95% CI for all-cause mortality stratified according to choline, betaine, or TMAO according to tertiles. Adjustments were made for individual traditional risk factors including age, gender, history of cardiovascular disease, history of heart failure, systolic blood pressure, LDL cholesterol, HDL cholesterol, smoking, as well as log-transformed body mass index, log-transformed hsCRP, Hb A1c, and log-transformed eGFR. We used category-free net reclassification improvement (NRI) and integrated discrimination improvement (IDI) to quantify improvement in model performance, and compared c-statistics between the 2 models with vs without TMAO. Statistical analyses were performed using SAS software (version 9.3, SAS Institute) and R software (version 3.1.2).
Results
Baseline characteristics of our study cohort of 1216 patients with T2DM undergoing elective diagnostic coronary angiography are shown in Table 1, stratified by tertiles of plasma TMAO. A significant proportion of patients had a history of cardiovascular disease, with nearly half of the patients having a history of prior MI and roughly a third of the patients having a history of revascularization via coronary artery bypass graft, and/or percutaneous intervention. Participants with increased Hb A1c levels tended to be younger and a higher proportion were female. In addition, individuals with increased Hb A1c were more likely to have lower HDL, higher hsCRP, higher choline, and be on either an angiotensin converting enzyme inhibitor or angiotensin receptor blocker (Table 1). Apart from correlations between TMAO and choline with eGFR, the majority of correlations between the 3 metabolites and laboratory measurements were weak (see Supplemental Table 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol63/issue1).
Baseline characteristics of diabetic cohort stratified according to TMAO tertiles.
| Total (n = 1216) | Tertile 1 (n = 401) | Tertile 2 (n = 414) | Tertile 3 (n = 401) | P value | |
|---|---|---|---|---|---|
| Age, years | 64.4 ± 10.2 | 61.1 ± 9.7 | 65.2 ± 9.9 | 66.9 ± 10.1 | <0.001 |
| Male, % | 58 | 64 | 57 | 53 | 0.004 |
| Hypertension, % | 79 | 74 | 76 | 88 | <0.001 |
| Smoking, % | 63 | 65 | 60 | 65 | 0.355 |
| History of MI, % | 47 | 43 | 46 | 51 | 0.067 |
| History of stroke, % | 8 | 5 | 7 | 12 | 0.002 |
| History of CABG,a % | 34 | 28 | 33 | 41 | <0.001 |
| History of PCI, % | 35 | 36 | 34 | 35 | 0.796 |
| LDL cholesterol, mg/dLb | 95 (76–115) | 96 (77–114) | 95 (79–118) | 93 (73–114) | 0.203 |
| HDL cholesterol, mg/dLb | 33 (27–40) | 35 (28–41) | 33 (28–39) | 32 (26–38) | <0.001 |
| hsCRP, mg/L | 3.3 (1.3–8.3) | 2.9 (1.2–8.2) | 3.2 (1.4–7.9) | 4.0 (1.4–8.8) | 0.155 |
| eGFR, mL · min−1 · (1.73 m2)−1 | 82 (62–94) | 92 (83–100) | 80 (65–93) | 61 (45–82) | <0.001 |
| ACE inhibitor/ARB, % | 59 | 55 | 60 | 63 | 0.038 |
| Statins, % | 64 | 67 | 64 | 62 | 0.335 |
| β-Blockers, % | 66 | 66 | 66 | 66 | 0.982 |
| Aspirin, % | 75 | 76 | 78 | 72 | 0.112 |
| Glucose-lowering drugs, % | 55 | 46 | 54 | 64 | <0.001 |
| TMAO, μmol/L | 4.4 (2.8–7.7) | 2.3 (1.7–2.8) | 4.4 (3.7–5.3) | 9.7 (7.8–14.9) | <0.001 |
| Choline, μmol/L | 10.6 (8.4–13.5) | 9.2 (7.7–11.1) | 10.4 (8.6–13.2) | 12.6 (10.0–16.2) | <0.001 |
| Betaine, μmol/L | 39 (30.7–49.6) | 37.6 (30.6–47) | 38.8 (30.1–49.7) | 40.7 (31.7–51.9) | 0.022 |
| Total (n = 1216) | Tertile 1 (n = 401) | Tertile 2 (n = 414) | Tertile 3 (n = 401) | P value | |
|---|---|---|---|---|---|
| Age, years | 64.4 ± 10.2 | 61.1 ± 9.7 | 65.2 ± 9.9 | 66.9 ± 10.1 | <0.001 |
| Male, % | 58 | 64 | 57 | 53 | 0.004 |
| Hypertension, % | 79 | 74 | 76 | 88 | <0.001 |
| Smoking, % | 63 | 65 | 60 | 65 | 0.355 |
| History of MI, % | 47 | 43 | 46 | 51 | 0.067 |
| History of stroke, % | 8 | 5 | 7 | 12 | 0.002 |
| History of CABG,a % | 34 | 28 | 33 | 41 | <0.001 |
| History of PCI, % | 35 | 36 | 34 | 35 | 0.796 |
| LDL cholesterol, mg/dLb | 95 (76–115) | 96 (77–114) | 95 (79–118) | 93 (73–114) | 0.203 |
| HDL cholesterol, mg/dLb | 33 (27–40) | 35 (28–41) | 33 (28–39) | 32 (26–38) | <0.001 |
| hsCRP, mg/L | 3.3 (1.3–8.3) | 2.9 (1.2–8.2) | 3.2 (1.4–7.9) | 4.0 (1.4–8.8) | 0.155 |
| eGFR, mL · min−1 · (1.73 m2)−1 | 82 (62–94) | 92 (83–100) | 80 (65–93) | 61 (45–82) | <0.001 |
| ACE inhibitor/ARB, % | 59 | 55 | 60 | 63 | 0.038 |
| Statins, % | 64 | 67 | 64 | 62 | 0.335 |
| β-Blockers, % | 66 | 66 | 66 | 66 | 0.982 |
| Aspirin, % | 75 | 76 | 78 | 72 | 0.112 |
| Glucose-lowering drugs, % | 55 | 46 | 54 | 64 | <0.001 |
| TMAO, μmol/L | 4.4 (2.8–7.7) | 2.3 (1.7–2.8) | 4.4 (3.7–5.3) | 9.7 (7.8–14.9) | <0.001 |
| Choline, μmol/L | 10.6 (8.4–13.5) | 9.2 (7.7–11.1) | 10.4 (8.6–13.2) | 12.6 (10.0–16.2) | <0.001 |
| Betaine, μmol/L | 39 (30.7–49.6) | 37.6 (30.6–47) | 38.8 (30.1–49.7) | 40.7 (31.7–51.9) | 0.022 |
CABG, coronary artery bypass graft; PCI, percutaneous coronary intervention; ACE, angiotensin converting enzyme; ARB, angiotensin II receptor blocker.
To convert mg/dL to mmol/L, multiply cholesterol by 0.02586.
Baseline characteristics of diabetic cohort stratified according to TMAO tertiles.
| Total (n = 1216) | Tertile 1 (n = 401) | Tertile 2 (n = 414) | Tertile 3 (n = 401) | P value | |
|---|---|---|---|---|---|
| Age, years | 64.4 ± 10.2 | 61.1 ± 9.7 | 65.2 ± 9.9 | 66.9 ± 10.1 | <0.001 |
| Male, % | 58 | 64 | 57 | 53 | 0.004 |
| Hypertension, % | 79 | 74 | 76 | 88 | <0.001 |
| Smoking, % | 63 | 65 | 60 | 65 | 0.355 |
| History of MI, % | 47 | 43 | 46 | 51 | 0.067 |
| History of stroke, % | 8 | 5 | 7 | 12 | 0.002 |
| History of CABG,a % | 34 | 28 | 33 | 41 | <0.001 |
| History of PCI, % | 35 | 36 | 34 | 35 | 0.796 |
| LDL cholesterol, mg/dLb | 95 (76–115) | 96 (77–114) | 95 (79–118) | 93 (73–114) | 0.203 |
| HDL cholesterol, mg/dLb | 33 (27–40) | 35 (28–41) | 33 (28–39) | 32 (26–38) | <0.001 |
| hsCRP, mg/L | 3.3 (1.3–8.3) | 2.9 (1.2–8.2) | 3.2 (1.4–7.9) | 4.0 (1.4–8.8) | 0.155 |
| eGFR, mL · min−1 · (1.73 m2)−1 | 82 (62–94) | 92 (83–100) | 80 (65–93) | 61 (45–82) | <0.001 |
| ACE inhibitor/ARB, % | 59 | 55 | 60 | 63 | 0.038 |
| Statins, % | 64 | 67 | 64 | 62 | 0.335 |
| β-Blockers, % | 66 | 66 | 66 | 66 | 0.982 |
| Aspirin, % | 75 | 76 | 78 | 72 | 0.112 |
| Glucose-lowering drugs, % | 55 | 46 | 54 | 64 | <0.001 |
| TMAO, μmol/L | 4.4 (2.8–7.7) | 2.3 (1.7–2.8) | 4.4 (3.7–5.3) | 9.7 (7.8–14.9) | <0.001 |
| Choline, μmol/L | 10.6 (8.4–13.5) | 9.2 (7.7–11.1) | 10.4 (8.6–13.2) | 12.6 (10.0–16.2) | <0.001 |
| Betaine, μmol/L | 39 (30.7–49.6) | 37.6 (30.6–47) | 38.8 (30.1–49.7) | 40.7 (31.7–51.9) | 0.022 |
| Total (n = 1216) | Tertile 1 (n = 401) | Tertile 2 (n = 414) | Tertile 3 (n = 401) | P value | |
|---|---|---|---|---|---|
| Age, years | 64.4 ± 10.2 | 61.1 ± 9.7 | 65.2 ± 9.9 | 66.9 ± 10.1 | <0.001 |
| Male, % | 58 | 64 | 57 | 53 | 0.004 |
| Hypertension, % | 79 | 74 | 76 | 88 | <0.001 |
| Smoking, % | 63 | 65 | 60 | 65 | 0.355 |
| History of MI, % | 47 | 43 | 46 | 51 | 0.067 |
| History of stroke, % | 8 | 5 | 7 | 12 | 0.002 |
| History of CABG,a % | 34 | 28 | 33 | 41 | <0.001 |
| History of PCI, % | 35 | 36 | 34 | 35 | 0.796 |
| LDL cholesterol, mg/dLb | 95 (76–115) | 96 (77–114) | 95 (79–118) | 93 (73–114) | 0.203 |
| HDL cholesterol, mg/dLb | 33 (27–40) | 35 (28–41) | 33 (28–39) | 32 (26–38) | <0.001 |
| hsCRP, mg/L | 3.3 (1.3–8.3) | 2.9 (1.2–8.2) | 3.2 (1.4–7.9) | 4.0 (1.4–8.8) | 0.155 |
| eGFR, mL · min−1 · (1.73 m2)−1 | 82 (62–94) | 92 (83–100) | 80 (65–93) | 61 (45–82) | <0.001 |
| ACE inhibitor/ARB, % | 59 | 55 | 60 | 63 | 0.038 |
| Statins, % | 64 | 67 | 64 | 62 | 0.335 |
| β-Blockers, % | 66 | 66 | 66 | 66 | 0.982 |
| Aspirin, % | 75 | 76 | 78 | 72 | 0.112 |
| Glucose-lowering drugs, % | 55 | 46 | 54 | 64 | <0.001 |
| TMAO, μmol/L | 4.4 (2.8–7.7) | 2.3 (1.7–2.8) | 4.4 (3.7–5.3) | 9.7 (7.8–14.9) | <0.001 |
| Choline, μmol/L | 10.6 (8.4–13.5) | 9.2 (7.7–11.1) | 10.4 (8.6–13.2) | 12.6 (10.0–16.2) | <0.001 |
| Betaine, μmol/L | 39 (30.7–49.6) | 37.6 (30.6–47) | 38.8 (30.1–49.7) | 40.7 (31.7–51.9) | 0.022 |
CABG, coronary artery bypass graft; PCI, percutaneous coronary intervention; ACE, angiotensin converting enzyme; ARB, angiotensin II receptor blocker.
To convert mg/dL to mmol/L, multiply cholesterol by 0.02586.
Relationship between plasma concentrations of TMAO, choline or betaine, and cardiovascular and mortality risks.a
| HR | Major adverse cardiac events at 3 years | All-cause mortality at 5 years | ||||
|---|---|---|---|---|---|---|
| TMAO, μmol/L | <3.2 | 3.2–6.3 | ≥6.3 | <3.2 | 3.2–6.3 | ≥6.3 |
| Unadjusted model | 1 | 1.95 (1.31–2.9)** | 3.03 (2.08–4.42)*** | 1 | 1.77 (1.19–2.64)** | 3.63 (2.53–5.21)*** |
| Adjusted model | 1 | 1.75 (1.15–2.66)** | 1.94 (1.23–3.05)** | 1 | 1.46 (0.96–2.22) | 1.85 (1.21–2.84)** |
| Events | 36/401 | 71/414 | 102/401 | 38/401 | 66/414 | 123/401 |
| Choline, μmol/L | <9.1 | 9.1–12.3 | ≥12.3 | <9.1 | 9.1–12.3 | ≥12.3 |
| Unadjusted model | 1 | 1.86 (1.28–2.71)** | 2.2 (1.52–3.17)*** | 1 | 1.81 (1.24–2.63)** | 2.64 (1.85–3.77)*** |
| Adjusted model | 1 | 1.55 (1.03–2.33)* | 1.38 (0.89–2.13) | 1 | 1.41 (0.95–2.09) | 1.36 (0.91–2.05) |
| Events | 42/401 | 78/412 | 89/403 | 43/401 | 77/412 | 107/403 |
| Betaine, μmol/L | <33.2 | 33.2–45.7 | ≥45.7 | <33.2 | 33.2–45.7 | ≥45.7 |
| Unadjusted model | 1 | 1.11 (0.78–1.59) | 1.62 (1.16–2.27)** | 1 | 1.3 (0.92–1.84) | 1.88 (1.35–2.63)** |
| Adjusted model | 1 | 1.08 (0.74–1.58) | 1.41 (0.97–2.03) | 1 | 1.32 (0.92–1.91) | 1.57 (1.09–2.26)* |
| Events | 56/400 | 65/415 | 88/401 | 55/400 | 74/415 | 98/401 |
| HR | Major adverse cardiac events at 3 years | All-cause mortality at 5 years | ||||
|---|---|---|---|---|---|---|
| TMAO, μmol/L | <3.2 | 3.2–6.3 | ≥6.3 | <3.2 | 3.2–6.3 | ≥6.3 |
| Unadjusted model | 1 | 1.95 (1.31–2.9)** | 3.03 (2.08–4.42)*** | 1 | 1.77 (1.19–2.64)** | 3.63 (2.53–5.21)*** |
| Adjusted model | 1 | 1.75 (1.15–2.66)** | 1.94 (1.23–3.05)** | 1 | 1.46 (0.96–2.22) | 1.85 (1.21–2.84)** |
| Events | 36/401 | 71/414 | 102/401 | 38/401 | 66/414 | 123/401 |
| Choline, μmol/L | <9.1 | 9.1–12.3 | ≥12.3 | <9.1 | 9.1–12.3 | ≥12.3 |
| Unadjusted model | 1 | 1.86 (1.28–2.71)** | 2.2 (1.52–3.17)*** | 1 | 1.81 (1.24–2.63)** | 2.64 (1.85–3.77)*** |
| Adjusted model | 1 | 1.55 (1.03–2.33)* | 1.38 (0.89–2.13) | 1 | 1.41 (0.95–2.09) | 1.36 (0.91–2.05) |
| Events | 42/401 | 78/412 | 89/403 | 43/401 | 77/412 | 107/403 |
| Betaine, μmol/L | <33.2 | 33.2–45.7 | ≥45.7 | <33.2 | 33.2–45.7 | ≥45.7 |
| Unadjusted model | 1 | 1.11 (0.78–1.59) | 1.62 (1.16–2.27)** | 1 | 1.3 (0.92–1.84) | 1.88 (1.35–2.63)** |
| Adjusted model | 1 | 1.08 (0.74–1.58) | 1.41 (0.97–2.03) | 1 | 1.32 (0.92–1.91) | 1.57 (1.09–2.26)* |
| Events | 56/400 | 65/415 | 88/401 | 55/400 | 74/415 | 98/401 |
Adjusted for traditional risk factors include age, sex, history of cardiovascular disease, systolic blood pressure, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, smoking, hsCRP, log-transformed Hb A1c, log-transformed eGFR, log-transformed body mass index, and history of heart failure;
P < 0.05,
P < 0.01,
P < 0.001.
Relationship between plasma concentrations of TMAO, choline or betaine, and cardiovascular and mortality risks.a
| HR | Major adverse cardiac events at 3 years | All-cause mortality at 5 years | ||||
|---|---|---|---|---|---|---|
| TMAO, μmol/L | <3.2 | 3.2–6.3 | ≥6.3 | <3.2 | 3.2–6.3 | ≥6.3 |
| Unadjusted model | 1 | 1.95 (1.31–2.9)** | 3.03 (2.08–4.42)*** | 1 | 1.77 (1.19–2.64)** | 3.63 (2.53–5.21)*** |
| Adjusted model | 1 | 1.75 (1.15–2.66)** | 1.94 (1.23–3.05)** | 1 | 1.46 (0.96–2.22) | 1.85 (1.21–2.84)** |
| Events | 36/401 | 71/414 | 102/401 | 38/401 | 66/414 | 123/401 |
| Choline, μmol/L | <9.1 | 9.1–12.3 | ≥12.3 | <9.1 | 9.1–12.3 | ≥12.3 |
| Unadjusted model | 1 | 1.86 (1.28–2.71)** | 2.2 (1.52–3.17)*** | 1 | 1.81 (1.24–2.63)** | 2.64 (1.85–3.77)*** |
| Adjusted model | 1 | 1.55 (1.03–2.33)* | 1.38 (0.89–2.13) | 1 | 1.41 (0.95–2.09) | 1.36 (0.91–2.05) |
| Events | 42/401 | 78/412 | 89/403 | 43/401 | 77/412 | 107/403 |
| Betaine, μmol/L | <33.2 | 33.2–45.7 | ≥45.7 | <33.2 | 33.2–45.7 | ≥45.7 |
| Unadjusted model | 1 | 1.11 (0.78–1.59) | 1.62 (1.16–2.27)** | 1 | 1.3 (0.92–1.84) | 1.88 (1.35–2.63)** |
| Adjusted model | 1 | 1.08 (0.74–1.58) | 1.41 (0.97–2.03) | 1 | 1.32 (0.92–1.91) | 1.57 (1.09–2.26)* |
| Events | 56/400 | 65/415 | 88/401 | 55/400 | 74/415 | 98/401 |
| HR | Major adverse cardiac events at 3 years | All-cause mortality at 5 years | ||||
|---|---|---|---|---|---|---|
| TMAO, μmol/L | <3.2 | 3.2–6.3 | ≥6.3 | <3.2 | 3.2–6.3 | ≥6.3 |
| Unadjusted model | 1 | 1.95 (1.31–2.9)** | 3.03 (2.08–4.42)*** | 1 | 1.77 (1.19–2.64)** | 3.63 (2.53–5.21)*** |
| Adjusted model | 1 | 1.75 (1.15–2.66)** | 1.94 (1.23–3.05)** | 1 | 1.46 (0.96–2.22) | 1.85 (1.21–2.84)** |
| Events | 36/401 | 71/414 | 102/401 | 38/401 | 66/414 | 123/401 |
| Choline, μmol/L | <9.1 | 9.1–12.3 | ≥12.3 | <9.1 | 9.1–12.3 | ≥12.3 |
| Unadjusted model | 1 | 1.86 (1.28–2.71)** | 2.2 (1.52–3.17)*** | 1 | 1.81 (1.24–2.63)** | 2.64 (1.85–3.77)*** |
| Adjusted model | 1 | 1.55 (1.03–2.33)* | 1.38 (0.89–2.13) | 1 | 1.41 (0.95–2.09) | 1.36 (0.91–2.05) |
| Events | 42/401 | 78/412 | 89/403 | 43/401 | 77/412 | 107/403 |
| Betaine, μmol/L | <33.2 | 33.2–45.7 | ≥45.7 | <33.2 | 33.2–45.7 | ≥45.7 |
| Unadjusted model | 1 | 1.11 (0.78–1.59) | 1.62 (1.16–2.27)** | 1 | 1.3 (0.92–1.84) | 1.88 (1.35–2.63)** |
| Adjusted model | 1 | 1.08 (0.74–1.58) | 1.41 (0.97–2.03) | 1 | 1.32 (0.92–1.91) | 1.57 (1.09–2.26)* |
| Events | 56/400 | 65/415 | 88/401 | 55/400 | 74/415 | 98/401 |
Adjusted for traditional risk factors include age, sex, history of cardiovascular disease, systolic blood pressure, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, smoking, hsCRP, log-transformed Hb A1c, log-transformed eGFR, log-transformed body mass index, and history of heart failure;
P < 0.05,
P < 0.01,
P < 0.001.
We first performed a cross-sectional comparison of TMAO concentrations between this study cohort with an independent set of 300 prospectively recruited, apparently healthy individuals without known cardiac diseases or T2DM from a health screening program at various locations across Cleveland, OH (see baseline characteristics in Supplemental Table 2 in the online Data Supplement). We observed that the median (IQR) plasma concentrations for TMAO [4.4 (2.8–7.7) μmol/L vs 3.4 (2.3–5.8) μmol/L, P < 0.001], choline [10.6 (8.4–13.5) μmol/L vs 8.2 (7.0–9.7) μmol/L, P < 0.001], and betaine concentrations [39 (30.7–49.6) μmol/L vs 37.7 (29.8–47.9) μmol/L, P = 0.028] were each significantly higher in patients with T2DM compared to the healthy control cohort (Fig. 1).
Comparisons in fasting TMAO (A), choline (B), and betaine (C) concentrations between individuals with T2DM (n = 1216) and healthy, nondiabetic controls (n = 300).
In our study cohort with 1216 T2DM patients, a total of 209 major adverse cardiac events occurred over the initial 3 years following enrollment, and 227 deaths occurred over the ensuing 5 years following enrollment. Kaplan–Meier analyses for both major adverse cardiac events and mortality outcomes are shown in Fig. 2, with plasma concentrations of TMAO, choline, and betaine each stratified by tertiles. We observed that increased fasting TMAO concentrations were associated with increased risk for 3-year major adverse cardiac events (tertiles 3 vs 1, HR = 3.03, 95% CI, 2.08–4.42, P < 0.001) and 5-year all-cause mortality (tertiles 3 vs 1, HR = 3.63, 95% CI, 2.53–5.21, P < 0.001). Similarly, increased fasting choline concentrations and betaine concentrations conferred increased mortality risks in both cohorts (see Supplemental Table 2 in the online Data Supplement). After adjustments for traditional risk factors, hsCRP, Hb A1c, history of cardiovascular disease, and eGFR, increased concentrations of fasting TMAO (adjusted HR = 2.07, 95% CI, 1.37–3.14, P < 0.01), choline (adjusted HR = 1.52, 95% CI, 1.03–2.26, P < 0.05), and betaine (adjusted HR = 1.59, 95% CI, 1.11–2.26, P < 0.05) all remained independently predictive of increased risk for all-cause mortality (see Supplemental Table 2 in the online Data Supplement). Cubic spline curves of HRs for major adverse cardiac events and all-cause mortality showed progressively increased risk with higher concentrations of TMAO, choline, and betaine (Fig. 3). Similar but somewhat attenuated dose-dependent relationships were observed between systemic choline and betaine concentrations and major adverse cardiac event risks in patients. Furthermore, the highest mortality risk cohorts for choline and betaine stratification were also confined to those with increased TMAO concentrations (see Fig. 1 in the online Data Supplement). When all 3 metabolites were included into the same adjusted model, the prognostic value of TMAO remained statistically significant (major adverse cardiac events: HR = 1.86, 95% CI, 1.19–2.92, P < 0.01; all-cause mortality: HR = 1.78, 95% CI, 1.15–2.73, P < 0.01). The inclusion of TMAO as a covariate resulted in a significant improvement in risk estimation over traditional risk factors for major adverse cardiac events [NRI, 25.8%, (P < 0.001); IDI, 2%, (P = 0.01)] and for all-cause mortality [NRI, 35.6%, (P < 0.001); IDI, 2%, (P = 0.02)], although neither saw a statistically significant increment in c-statistics (major adverse cardiac events: AUC from 69.6% to 71.2%, P = 0.11; all-cause mortality: from 74.4% to 75.3%, P = 0.186).
Kaplan–Meier survival analysis of 3-year risk of major adverse cardiac events and all-cause mortality stratified according to fasting TMAO (A, D), choline (B, E), and betaine (C, F) tertiles.
Cubic spline curves of HRs for major adverse clinical events (A–C) (death, nonfatal MI, and stroke) at 3 years and all-cause mortality at 5 years (D–F) with fasting concentrations of TMAO (A, B), choline (C, D), and betaine (E, F).
We further investigated the association between phosphatidylcholine metabolite (i.e., TMAO, choline and betaine) concentrations with glycemic status, as quantified by Hb A1c levels, within our T2DM cohort. Overall, we observed a modest but statistically significant correlation between choline and Hb A1c (r = 0.08, P = 0.004), while TMAO and Hb A1c (r = 0.02, P = 0.527) and betaine and Hb A1c (r = −0.03, P = 0.36) did not show significant correlations (see Table 1 in the online Data Supplement). The T2DM patients were further divided according to median phosphatidylcholine metabolite concentrations and further stratified according to glycemic status (above and below Hb A1c of 6.5%). We observed that stratification by TMAO, choline, and betaine concentrations conferred stronger major adverse cardiac event risk and mortality risk compared to glycemic status (Fig. 4). These results were similar when stratified by Hb A1c cutoff of 7.5% (data not shown). Specifically, T2DM patients with high concentrations of phosphatidylcholine metabolites but good glycemic control had higher mortality risk than those with low concentrations of phosphatidylcholine metabolites but poor glycemic control. Compared to those with low phosphatidylcholine metabolite concentrations and good glycemic control, T2DM patients with increased fasting phosphatidylcholine metabolite concentrations and with concomitantly poor glycemic control had the highest mortality risk [i.e., TMAO (HR = 3.2, 95% CI, 1.88–5.44, P < 0.001); betaine (HR = 1.67, 95% CI, 1.08–2.6, P = 0.021); choline (HR = 2.14, 95% CI, 1.34–3.42, P = 0.001)].
Kaplan–Meier survival analysis of 3-year risk of major adverse cardiac events (A–C) and all-cause mortality (D–F) stratified according to median fasting TMAO (4.4 μmol/L), choline (10.6 μmol/L), and betaine (39 μmol/L) concentrations with Hb A1c levels (6.5%).
Discussion
In this study, we tested the hypothesis that plasma concentrations of the gut microbe-generated metabolite TMAO and 2 of its nutrient precursors, choline and betaine, were associated with increased long-term mortality risk in patients with T2DM. Several key findings were noted in this analysis. First, plasma TMAO, choline, and betaine were each dose-dependently associated with greater 3-year major adverse cardiac event risk and 5-year mortality risk among stable patients with T2DM. Further, the increased 3-year major adverse cardiac event risk remained robust in the TMAO and betaine groups after adjusting for traditional risk factors, inflammation (hsCRP), glycemic status (Hb A1c), and history of cardiovascular disease and kidney function (eGFR). Moreover, both TMAO and betaine also remained significant for 5-year mortality risk after adjustments in the fully loaded model. Second, we confirmed previous reports (20, 21) that suggested patients with T2DM had higher TMAO, choline, and betaine concentrations than apparently healthy controls. Third, increased fasting TMAO, choline, and betaine concentrations were observed to be associated with worse prognosis regardless of glycemic status. Fourth, we observe that the prognostic value of increased choline and betaine are only observed amongst those with increased TMAO.
Participation of the gut microbiome in the pathogenesis of cardiometabolic diseases such as obesity, T2DM, and cardiovascular disease is emerging as an exciting and recent realization (5, 22). Studies assessing microbial composition have begun to identify T2DM enriched microbiota in different populations (20, 21). Similarly, specific gut microbial taxa appear to be enriched within individuals having cardiovascular disease (23). Furthermore, microbial transplantation studies reveal that impaired glycemic control and both atherosclerosis susceptibility and TMAO production are transmissible traits (10). Various bacterial enzyme complexes have recently been discovered that can generate the precursor to TMAO, TMA (13, 24, 25). Interestingly, many members of the Desulfovibrio genus, from which the first microbial choline TMA lyase enzyme was reported (26), represent a choline-degrading genus that can liberate TMA from dietary precursors. This sulfate-reducing bacterium has also been identified as a diabetogenic-associated taxon (21). In addition, common intestinal residents such as the Clostridium genus have been associated with both T2DM and increased TMAO cohorts (11, 21). However, these associations have not proven to be consistent in all populations (20). Nevertheless, it is tempting to speculate on the possibility of microbial “gut signatures” that may confer altered insulin sensitivity and TMAO production potential in the host, and thus correspond to specific disease states such as T2DM and cardiovascular risk.
Several groups have observed associations between choline and betaine with the prevalence of T2DM (17, 27–29). Moreover, many studies have observed increased concentrations of choline and betaine associated with the increased incidence of acute coronary syndromes, as well as increased blood lipid concentrations (14, 17, 30, 31). These studies suggest a potential link between choline and betaine with T2DM and cardiovascular health. However, the evidence is muddied by the fact that some epidemiological studies that predicted choline and betaine intake based upon food questionnaire data show only limited associations with cardiovascular morbidity that failed to remain significant after adjustments for comorbidities (32). Interestingly, in our study, despite the fact that higher plasma betaine concentrations predicted increased adverse prognosis, the average value of betaine was lower in the higher Hb A1c cohort. Betaine is known to play an important role in cell volume regulation and the reduction in homocysteine through remethylation (3). However, the relationships between betaine concentrations and cardiovascular risks are still unknown. Studies on betaine as a marker for cardiovascular disease, T2DM, or metabolic syndrome have resulted in widely contrasting marker levels in association with these disease processes (27, 28, 33). However, in general, these studies suggest that decreased betaine, whether as a result of deficiency or increased excretion, may be linked to cardiometabolic disease. The mechanisms for observed reduction in betaine concentrations in the setting of T2DM are unknown, although hypotheses may include increased betaine–homocysteine methyltransferase activity or betaine efflux dysregulation (17, 34).
Indeed, choline and betaine have been linked to changes in insulin resistance with increased choline bacterial conversion to TMA and TMAO associated with impaired glucose homeostasis in a nonalcoholic fatty liver animal model (4). Recent studies report that dietary TMAO enhances the impaired glucose tolerance observed in mice fed a high fat diet (35). Moreover, manipulation of TMAO concentrations in mice through inhibition in host flavin monooxygenase 3, the key host enzyme responsible for converting TMA into TMAO (12), demonstrates the pathway exerts broad effects on glucose metabolism (12, 36). Thus, some studies suggest mechanistic links between the TMAO pathway and T2DM. Consistent with this, in several clinical cohorts, including individuals undergoing elective diagnostic cardiac catheterization, as well as among patients with chronic systolic heart failure, we have observed in cross-section an increased prevalence of T2DM among individuals with increased TMAO (9, 14, 37).
In this study, while individuals with T2DM had higher TMAO, choline, and betaine concentrations relative to controls, we did not observe a strong relationship between the concentrations of TMAO, choline, or betaine and glycemic status in this T2DM population. Although choline was modestly correlated with Hb A1c levels, TMAO and betaine were not. Nonetheless, evidence of associations between increased TMAO, choline, and betaine concentrations and increased mortality across glycemic status suggest that these metabolites, especially TMAO, may serve as an important risk marker for adverse prognosis in T2DM, and be increased in T2DM by a common metabolic alteration. Furthermore, the dependence of TMAO on gut microbiota composition and its dietary relationship with choline and betaine lead us to believe that gut microbial processes generating TMAO from choline or betaine may represent a potential pathogenic link with both cardiovascular disease development and T2DM status. Whether targeting the gut microbial TMAO pathway pharmacologically can impact cardiovascular disease (and T2DM) risks in patients is not yet known. However, it is noteworthy that in recent studies, nonlethal pharmacological targeting of gut microbial TMA lyase activity was effective at both inhibiting TMA and TMAO production in vivo, and attenuating diet induced atherosclerosis in an animal model (38). Moreover, the TMA lyase inhibitor used was also found to be a natural product present in abundance in some cold-pressed extra virgin olive oils and grape seed oils (38). Consumption of extra virgin olive oil is a critical component of the Mediterranean diet, and adherence to a Mediterranean diet has been associated with both reduced concentrations of TMAO (39) and reduction in risk for both cardiovascular disease and T2DM development (40). Further studies on the relationship between the gut microbial TMAO pathway, and both T2DM and cardiovascular risks are warranted.
STUDY LIMITATIONS
As a single-center observational cohort, these findings need to be replicated in other validation cohorts. Furthermore, we do not have adequate information regarding the type and/or drug dosing of various glucose-lowering agents used and some drugs may be temporarily withheld or reduced in dosages in the setting of scheduled elective diagnostic coronary angiography. We also do not have information regarding dietary patterns, duration of diabetes/insulin use, or different types and severity of diabetic complications associated with microvascular dysfunction (neuropathy, nephropathy, and retinopathy). Similarly, no information was available regarding causes of death or nonprescriptive dietary supplements that may have been used by patients at the time of study. Regardless, the present studies suggest that further investigations into the role of modulating gut microbiota to modify cardiovascular risks in patients with T2DM are of interest in this vulnerable patient population.
Conclusion
The plasma concentration of TMAO, a proatherogenic compound formed by gut microbe–dependent metabolism of choline, is increased in individuals with T2DM and associated with increased risk of major adverse cardiac events and mortality in patients with T2DM independent of glycemic status.
5 Nonstandard abbreviations
- T2DM
type 2 diabetes mellitus
- TMAO
trimethylamine N-oxide
- TMA
trimethylamine
- Hb A1c
glycohemoglobin
- hsCRP
high-sensitivity C-reactive protein
- eGFR
estimated glomerular filtration rate
- MI
myocardial infarction
- IQR
interquartile range
- HR
hazard ratio
- NRI
net reclassification improvement
- IDI
integrated discrimination improvement.
Author Contributions:All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest:Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: None declared.
Consultant or Advisory Role: S.L. Hazen, Esperion and P&G.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: Cleveland Clinic Clinical Research Unit of the Case Western Reserve University CTSA (UL1TR 000439) supporting the GeneBank study. Mass spectrometry studies were performed on instruments housed in a facility supported in part by the Center of Innovations Award by AB Sciex. W.H.W. Tang, NIH grant R01HL103931; Z. Wang, American Heart Association Scientist Development Grant and NIH 1R01HL130819; S.L. Hazen, AstraZeneca, P&G, Pfizer Inc., Roche, Takeda, National Institutes of Health and the Office of Dietary Supplements (R01HL103866, P20HL113452, and R01DK106000), NIH grants P01HL076491 and P01HL098055, the Center of Innovations Award by AB Sciex, and the Leonard Krieger endowment.
Expert Testimony: None declared.
Patents: Z. Wang, pending patents held by the Cleveland Clinic relating to cardiovascular and inflammation diagnostics; S.L. Hazen, Treatment and Prevention of Cardiovascular Disease and Thrombosis, US patent no. 9,168,233. Z. Wang and S.L. Hazen are named as coinventors on pending and issued patents held by the Cleveland Clinic relating to cardiovascular diagnostics and therapeutics. S.L. Hazen reports having the right to receive royalty payments for inventions or discoveries related to cardiovascular diagnostics or therapeutics from the Cleveland Heart Lab., Siemens, Esperion, Frantz Biomarkers, LLC.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, and final approval of manuscript.
References
