Abstract
Background:
Effects of animal source foods (ASF) on atherosclerotic cardiovascular disease (ASCVD) and underlying mechanisms remain controversial. We investigated prospective associations of different ASF with incident ASCVD and potential mediation by gut microbiota-generated trimethylamine N-oxide, its L-carnitine-derived intermediates γ-butyrobetaine and crotonobetaine, and traditional ASCVD risk pathways.
Methods:
Among 3931 participants from a community-based US cohort aged 65+ years, ASF intakes and trimethylamine N-oxide-related metabolites were measured serially over time. Incident ASCVD (myocardial infarction, fatal coronary heart disease, stroke, other atherosclerotic death) was adjudicated over 12.5 years median follow-up. Cox proportional hazards models with time-varying exposures and covariates examined ASF-ASCVD associations; and additive hazard models, mediation proportions by different risk pathways.
Results:
After multivariable-adjustment, higher intakes of unprocessed red meat, total meat, and total ASF associated with higher ASCVD risk, with hazard ratios (95% CI) per interquintile range of 1.15 (1.01–1.30), 1.22 (1.07–1.39), and 1.18 (1.03–1.34), respectively. Trimethylamine N-oxide-related metabolites together significantly mediated these associations, with mediation proportions (95% CI) of 10.6% (1.0–114.5), 7.8% (1.0–32.7), and 9.2% (2.2–44.5), respectively. Processed meat intake associated with a nonsignificant trend toward higher ASCVD (1.11 [0.98–1.25]); intakes of fish, poultry, and eggs were not significantly associated. Among other risk pathways, blood glucose, insulin, and C-reactive protein, but not blood pressure or blood cholesterol, each significantly mediated the total meat-ASCVD association.
Conclusions:
In this large, community-based cohort, higher meat intake associated with incident ASCVD, partly mediated by microbiota-derived metabolites of L-carnitine, abundant in red meat. These novel findings support biochemical links between dietary meat, gut microbiome pathways, and ASCVD.
Graphical Abstract
Highlights
•
In a community-based cohort of older US adults aged ≥65 years, higher intakes of unprocessed red meat, total meat (unprocessed red meat plus processed meat), and total animal source foods were prospectively associated with a higher incidence of atherosclerotic cardiovascular disease during a median follow-up of 12.5 years.
•
These associations were partly mediated (8%–11% of excess risk) by plasma levels of gut microbiota-generated metabolites, including trimethylamine N-oxide and its 2 intermediates derived from L-carnitine, abundant in red meat.
•
The higher risk of atherosclerotic cardiovascular disease associated with meat intake was also partly mediated by glucose-insulin homeostasis and systematic inflammation but not blood pressure or blood cholesterol levels.
•
Intakes of fish, poultry, and eggs were not significantly associated with atherosclerotic cardiovascular disease.
Animal source foods (ASFs), including unprocessed red meat, processed meat, fish, poultry, and eggs, are major components of many diets. The impact of these different foods on atherosclerotic cardiovascular disease (ASCVD) have been widely studied but remain controversial. Evidence is particularly sparse among older adults, the age group at the highest risk for ASCVD and in whom adequate intakes of high-quality protein, which is rich in ASF, appears important to offset aging-related losses of muscle mass and strength.1–3
The resulting controversies are exacerbated by poorly understood potential mechanisms underlying these associations. A historical focus on saturated fat, for example, has been tempered by evidence that its health effects vary according to the food source,4 suggesting relevance of other compounds in ASF. Growing evidence highlights newly discovered, gut microbiota-generated metabolites of ASF.5–7 These include trimethylamine N-oxide (TMAO), generated by microbial metabolism of dietary L-carnitine (abundant almost exclusively in red meat) and choline (present in a variety of ASF). Gut microbial metabolism of L-carnitine also generates 2 intermediates γ-butyrobetaine and crotonobetaine, each of which can then be further converted to TMAO (Figure 1). In experiments, TMAO promotes macrophage foam cell formation,8 vascular inflammation and inflammasome activation,9–12 endothelial dysfunction,13 platelet hyperreactivity and thrombosis,14,15 and decreases reverse cholesterol transport.16 In large clinical samples of patients with prevalent diseases, although not other small studies with less robust designs,17,18 higher plasma levels of TMAO were associated with higher risk of ASCVD and total mortality.8,16,19–23 It has been hypothesized that TMAO and its gut microbiota-generated intermediates may partly mediate the effects of consumption of ASF on ASCVD.24 However, no research has assessed this hypothesis. Investigation of such mediation would help advance understanding of potential mechanisms linking these ASF to ASCVD, as well as reasons for heterogeneous associations with ASCVD of different ASF.
To address these important research gaps, we investigated the associations of different ASF with incidence of ASCVD in a prospective, community-based cohort of older adults. We further evaluated the extent to which plasma levels of TMAO, γ-butyrobetaine, and crotonobetaine (referred to hereafter as TMAO-related metabolites) jointly mediated the identified associations.
Materials and Methods
Because of the sensitive nature of the data collected for this study, requests to access the data set from qualified researchers trained in human subject confidentiality protocols may be sent to the Cardiovascular Health Study Collaborative Health Studies Coordinating Center at [email protected].
Study Population
The CHS (Cardiovascular Health Study) is a multicenter, community-based, prospective cohort study designed to investigate risk factors for coronary heart disease and stroke in older adults. The study design and participant recruitment have been described.25,26 Briefly, in 1989 to 1990, 5201 noninstitutionalized adults aged ≥65 years were recruited from random samples of Medicare eligibility lists in 4 US communities. To enrich minority recruitment, an additional 687 Black participants were recruited in 1992 to 1993 using similar methods, resulting in 5888 total participants. Trained personnel assessed participants’ demographic characteristics, lifestyle, medical history, and other health related phenotypes during annual in-clinic exams with intervening phone interviews every 6 months through 1999. Thereafter, participants were contacted every 6 months by phone for follow-up through June 2015. Follow-up for vital status was nearly 100% complete. The study was approved by the institutional review board of each participating university. All participants provided written informed consent.
After excluding participants without joint assessments of diet and TMAO, with extreme reported energy intake (<500 or >5000 kcal/d), and with prevalent CVD (myocardial infarction, stroke, angina, and coronary revascularization) at the time of their first joint assessment of diet and TMAO, a total of 3931 participants were included in this investigation (Figure S1). Compared with included individuals, excluded individuals were slightly more likely to be male, non-White, older, less educated, and less healthy by self-report (Table S1).
Assessment of Dietary Habits (Exposures)
We focused on foods with significant associations with ASCVD in prior meta-analyses of generally middle-aged populations, including processed meat, unprocessed red meat, and fish.28–33 We hypothesized that processed meat and unprocessed red meat consumption would be positively associated with incidence of ASCVD and that these associations would be partly mediated by plasma levels of TMAO-related metabolites; and that fish consumption would be inversely associated with ASCVD, and that its association would become stronger (ie, more protective) after accounting for plasma levels of TMAO-related metabolites. In exploratory analyses, we examined the associations of secondary dietary exposures including total meat (ie, unprocessed red meat plus processed meat), poultry, eggs, and total of these ASF with ASCVD. Dairy foods were not included given these are not appreciable dietary sources of TMAO precursors.
Usual dietary habits over the past year were assessed in 1989 to 1990 using a validated 99-item picture-sort food frequency questionnaire (FFQ) adapted from the National Cancer Institute,34,35 and again in 1995 to 1996 using a validated Willett semiquantitative FFQ.36–38 The Pearson correlation coefficients between the Willett FFQ and two 1-week dietary records ranged from 0.56 to 0.83 for individual food groups of meats, eggs, and fish.37 For each FFQ, participants were asked to indicate how often, on average, they had eaten given amounts of various foods during the past year. The picture-sort FFQ used a 5-category frequency of intake ranging from never to almost every day or at least 5 times per week, based on medium portion sizes. The Willett FFQ used a 10-category frequency of intake ranging from never or less than once per month to 6+ per day, with defined standard portion sizes. Frequencies of intake were converted to servings/d using the midpoint of the relevant response category.39,40 Food intakes were adjusted for total energy using the residual method.41
Assessment of Plasma TMAO-Related Metabolites (Mediators)
We primarily focused on the joint mediation by TMAO and its 2 intermediate gut microbiota-dependent metabolites derived from dietary L-carnitine, γ-butyrobetaine, and crotonobetaine. We also explored path-specific mediated associations by each TMAO-related metabolite and mediation by plasma levels of each nutrient precursor (ie, choline, betaine, and carnitine). Measurements were performed using stored frozen (−80 °C) fasting blood samples collected at enrollment (1989–1990 or 1992–1993) and again in 1996 to 1997. Each biomarker was quantified using its deuterium-isotopologue as internal standard via a stable-isotope dilution assay coupled with high-performance liquid chromatography, with online electrospray ionization tandem mass spectrometry on a Shimadzu 8050 mass spectrometer. All laboratory measurements were performed at the Cleveland Clinic Lerner Research Institute, with laboratory coefficient of variation <10% for each metabolite, as previously described.7
Assessment of Traditional ASCVD Risk Factors and Other Covariates
At each in-clinic exam, information on sociodemographics, lifestyle, anthropometrics, medical history, medications (including antibiotic use in the past 2 weeks), and other risk factors were assessed by trained personnel using standardized questionnaires and physical examination.25 Physical activity (excluding chores, kcal/wk) was assessed by a modified Minnesota Leisure Time Activities Questionnaire.42,43 Information was collected on alcohol intake, including usual frequency and types of alcoholic beverages (wine, beer, and liquor), smoking status (never, former, or current; lifetime pack-years), and self-perceived general health (excellent, very good, good, fair, and poor). Anthropometrics were directly measured, as were 2 seated resting blood pressure measurements. Fasting glucose, total cholesterol, HDL (high-density lipoprotein) cholesterol, and triglyceride levels were measured from collected blood samples using standardized methods; and LDL (low-density lipoprotein) cholesterol level calculated using the Friedewald formula excluding patients with hypertriglyceridemia. CRP (C-reactive protein) was measured using a high-sensitivity ELISA.44 Cystatin-C and creatinine were measured and used to calculate estimated glomerular filtration rate (eGFR) using the Chronic Kidney Disease Epidemiology Collaboration equation.45–47 Diabetes was defined by treatment with oral hypoglycemic agents or insulin, fasting plasma glucose ≥126 mg/dL, or 2-hour postoral glucose challenge ≥200 mg/dL.48
Assessment of ASCVD (Outcome)
The primary outcome was incident ASCVD, defined as a composite of first definite or probable myocardial infarction, fatal coronary heart disease (CHD), stroke (excluding transient ischemic attack), or other atherosclerotic death. Potential ASCVD events were identified during annual examinations and interim telephone interviews.49 All ASCVD events were adjudicated continuously from baseline through June 2015 by centralized committees based on information from interviews, medical records, physician questionnaires, death certificates, medical examiner forms, Health Care Financing Administration hospitalizations, and available brain imaging.49,50 The detailed methods for follow-up and classification of events have been published.49,50 Briefly, myocardial infarction was classified based on chest pain, cardiac enzymes, and ECG findings. Fatal events with suspected coronary cause not meeting criteria for myocardial infarction were classified as fatal CHD if occurring within 72 hours of chest pain or with an antecedent history of CHD.40 Stroke was defined as neurological deficit of rapid onset lasting longer than 24 hours unless death supervened or as a subarachnoid hemorrhage.
Statistical Analysis
Cox proportional hazards models with time-varying exposures and covariates assessed the association (hazard ratio [HR]) between each dietary exposure and incidence of ASCVD (ie, main dietary association). Time at risk was calculated from the first joint availability of diet and TMAO measures (ie, time zero) to the occurrence of ASCVD, death due to non-ASCVD reasons, or last study contact, whichever occurred first (Figure S2). The proportional hazards assumption was examined using a test based on Schoenfeld residuals.51 Given that 2 covariates (sex, self-perceived health status) violated the assumption, we used risk-set stratified Cox models for these 2 covariates in all multivariable-adjusted analyses.
To leverage serial dietary measures, reduce exposure misclassification, and obtain estimates of long-term dietary intake, we assessed time-varying cumulative averages of dietary consumption (ie, cumulative updating)41: dietary measures at or before the first TMAO measure in 1989 to 1990 were related to ASCVD risk until the time point of the second TMAO measure in 1996 to 1997, and the average of serial dietary measurements at or before 1996 to 1997 was related to subsequent risk until 2015. For participants with only one dietary measure, that dietary measure was carried forward. Dietary exposures were analyzed as linear terms measured in units of the difference between the midpoints of the first and fifth quintiles (interquintile range [IQR]). In sensitivity analyses, we used restricted cubic splines to explore potential nonlinear associations and used the most recent intake (ie, simple updating) instead of cumulative updating for time-varying dietary exposures.
We adjusted for prespecified covariates including age, sex, race, study site, education, and household income, and time-varying smoking status, alcohol intake, physical activity, self-perceived health status, antibiotic use, and dietary habits, including intakes of total energy, fruits, vegetables, dietary fiber, total dairy, and mutual adjustment for the other ASF. In sensitivity analyses, we further adjusted for traditional CVD risk factors that may be intermediate outcomes on the causal pathway between diet and ASCVD, including body mass index, waist circumference, diabetes, systolic blood pressure, diastolic blood pressure, LDL cholesterol, HDL cholesterol, triglycerides, CRP, antihypertensive medication use, and lipid-lowering medication use. In sensitivity analysis, we further adjusted for eGFR which could be both a confounder and mediator (ie, on the causal pathway) of TMAO’s effects on ASCVD.52 TMAO is renally cleared,7 which could make eGFR a confounder; and TMAO also experimentally causes renal fibrosis and dysfunction,53,54 which could make eGFR a mediator of the association with ASCVD. In this context, including eGFR in the primary model would be subject to overadjustment. All time-varying covariates were updated at the time of TMAO updating using the most recent measure. Covariates with missing values were imputed using single imputation via best-subset regression; previous studies in CHS have documented minimal differences in results using this approach compared with multiple imputation.55
We used additive hazard models to perform causal mediation analyses.56–58 The 3 TMAO-related metabolites (TMAO, γ-butyrobetaine, and crotonobetaine) were analyzed as time-varying linear variables. Simple updating was used for these mediators to ensure that mediators were measured no earlier than the measurement of dietary exposures (Figure S2). The associations between each dietary exposure and ASCVD (measured by rate difference) were decomposed into those independent of and mediated via the 3 TMAO-related metabolites based on the conceptual diagram shown in Figure 2. Mediation proportions were defined as the mediated association/|(independent association+mediated association)|. A detailed description of calculations of independent and mediated associations was included in the Supplemental Material. Figure S3 (study design flowchart) summarizes all main analyses performed in the study. Given that animal feeding studies have established the causal interconversions between these metabolites but not the precise order of the pathways,6,27 we evaluated alternative conceptual diagrams changing the sequence of the 3 mediators, and findings were not appreciably changed (data not shown). We also explored and compared mediation proportions for traditional ASCVD risk factors.
We explored effect modification by baseline renal function (eGFR <60 versus ≥60 mL/min per 1.73 m2) for main dietary associations, based on recent findings suggesting renal function could be an effect modifier of the TMAO-ASCVD association,52 using multiplicative interaction terms between each dietary exposure and eGFR. In post hoc exploratory analyses, we similarly explored effect modification by age (≥ versus <median, 72 years), sex, race and ethnicity (White versus non-White), education level (<high school, high school, some college, or college graduate), and smoking status (never smoked, former smoker, or current smoker); with Bonferroni correction for testing of these exploratory interactions (5 interaction variables×7 dietary exposures =35 comparisons; corrected threshold of significance: 0.05/35=0.0014). Analyses were performed using Stata version 14.2 (StataCorp) and R version 4.0.3 (The R Foundation). Statistical significance for main dietary associations was assessed using a 2-sided α=0.05. Statistical significance of mediation was assessed by the 95% CIs.
Results
Participant Characteristics
Among participants at baseline, mean (SD) age was 72.9 (5.5) years, most were female (63.5%), and 12.0% were non-White (Table 1). Educational attainment ranged from <high school (25.9%) to college graduates (21.4%). About 20% of participants had diabetes, 40% were on antihypertensive mediations, and 3% had taken antibiotics in the previous 2 weeks. Participants with higher unprocessed red meat intake were more likely to be male, current smokers, less educated, physically inactive, and have prevalent diabetes and have lower intakes of fruits, vegetables, and dietary fiber. Patterns of participant characteristics across quintiles of processed meat intake were similar (Table S2). Opposing patterns were observed for fish intake, with participants having higher intake being more likely female, never or former smokers, more educated, and to have higher intakes of fruits, vegetables, and dietary fiber (Table S3).
| Q1 | Q2 | Q3 | Q4 | Q5 | Total | |
|---|---|---|---|---|---|---|
| N | 778 | 778 | 778 | 778 | 777 | 3889* |
| Range, servings/d | (0.01–0.18) | (0.18–0.30) | (0.30–0.47) | (0.47–0.69) | (0.69–4.49) | (0.01–4.49) |
| Sociodemographic factors | ||||||
| Age, y | 72.9±5.5 | 72.7±5.4 | 73.2±6.0 | 73.0±5.7 | 72.5±5.4 | 72.9±5.6 |
| Male | 233 (29.9) | 260 (33.4) | 296 (38.0) | 336 (43.2) | 294 (37.8) | 1419 (36.5) |
| Race | ||||||
| White | 673 (86.5) | 672 (86.4) | 684 (87.9) | 692 (88.9) | 700 (90.1) | 3421 (88.0) |
| Non-White | 105 (13.5) | 106 (13.6) | 94 (12.1) | 86 (11.1) | 77 (9.9) | 468 (12.0) |
| Income | ||||||
| ≤$11 999 | 172 (22.1) | 179 (23.0) | 168 (21.6) | 163 (21.0) | 188 (24.2) | 870 (22.4) |
| $12 000–24 999 | 259 (33.3) | 264 (33.9) | 312 (40.1) | 299 (38.4) | 286 (36.8) | 1420 (36.5) |
| $25 000 to $49 999 | 218 (28.0) | 231 (29.7) | 200 (25.7) | 223 (28.7) | 195 (25.1) | 1067 (27.4) |
| ≥$50 000 | 129 (16.6) | 104 (13.4) | 98 (12.6) | 93 (12.0) | 108 (13.9) | 532 (13.7) |
| Education | ||||||
| <High school | 180 (23.1) | 181 (23.3) | 206 (26.5) | 211 (27.1) | 230 (29.6) | 1008 (25.9) |
| High school | 226 (29.0) | 236 (30.3) | 229 (29.4) | 221 (28.4) | 223 (28.7) | 1135 (29.2) |
| Some college | 180 (23.1) | 186 (23.9) | 183 (23.5) | 193 (24.8) | 173 (22.3) | 915 (23.5) |
| College graduate | 192 (24.7) | 175 (22.5) | 160 (20.6) | 153 (19.7) | 151 (19.4) | 831 (21.4) |
| Lifestyle factors | ||||||
| Smoking | ||||||
| Never | 386 (49.6) | 359 (46.1) | 390 (50.1) | 359 (46.1) | 357 (45.9) | 1851 (47.6) |
| Former | 324 (41.6) | 330 (42.4) | 288 (37.0) | 321 (41.3) | 312 (40.2) | 1575 (40.5) |
| Current | 68 (8.7) | 89 (11.4) | 100 (12.9) | 98 (12.6) | 108 (13.9) | 463 (11.9) |
| Alcohol, drinks/wk | 0.02 (0–1.25) | 0.04 (0–1.50) | 0.02 (0–1.25) | 0.02 (0–1.50) | 0 (0–1.00) | 0.02 (0–1.25) |
| Physical activity, kcal/wk | 699 (158–1643) | 690 (205–1530) | 600 (140–1650) | 623 (158–1470) | 496 (0–1328) | 622 (135–1530) |
| General health | ||||||
| Excellent | 125 (16.1) | 114 (14.7) | 122 (15.7) | 102 (13.1) | 112 (14.4) | 575 (14.8) |
| Very good | 225 (28.9) | 213 (27.4) | 216 (27.8) | 223 (28.7) | 202 (26.0) | 1079 (27.7) |
| Good | 296 (38.0) | 298 (38.3) | 308 (39.6) | 314 (40.4) | 291 (37.5) | 1507 (38.8) |
| Fair | 119 (15.3) | 140 (18.0) | 120 (15.4) | 120 (15.4) | 153 (19.7) | 652 (16.8) |
| Poor | 13 (1.7) | 13 (1.7) | 12 (1.5) | 19 (2.4) | 19 (2.4) | 76 (2.0) |
| Body mass index, kg/m2 | 26.0±4.5 | 26.2±4.6 | 26.9±4.6 | 27.0±4.7 | 27.1±4.9 | 26.6±4.7 |
| Medical history | ||||||
| Antihypertensive medication use | 302 (38.8) | 321 (41.3) | 313 (40.2) | 316 (40.6) | 319 (41.1) | 1571 (40.4) |
| Lipid-lowering medication use | 42 (5.4) | 48 (6.2) | 38 (4.9) | 32 (4.1) | 25 (3.2) | 185 (4.8) |
| Oral hypoglycemic agents or insulin | 37 (4.8) | 59 (7.6) | 38 (4.9) | 60 (7.7) | 57 (7.3) | 251 (6.5) |
| Antibiotic use (prior 2 wk) | 28 (3.6) | 18 (2.3) | 19 (2.4) | 24 (3.1) | 23 (3.0) | 112 (2.9) |
| Diabetes | 134 (17.2) | 153 (19.7) | 148 (19.0) | 171 (22.0) | 183 (23.6) | 789 (20.3) |
| Diet | ||||||
| Fruits, servings/d | 2.5±1.1 | 2.3±1.0 | 2.2±1.1 | 2.1±1.0 | 1.9±1.0 | 2.2±1.1 |
| Vegetables, servings/d | 2.9±1.4 | 2.6±1.2 | 2.6±1.2 | 2.5±1.2 | 2.5±1.2 | 2.6±1.3 |
| Dietary fiber, g/d | 33.6±10.9 | 31.0±10.0 | 29.5±9.5 | 28.7±9.0 | 26.6±9.2 | 29.9±10.0 |
| Dairy products, servings/d | 1.4±0.7 | 1.4±0.6 | 1.3±0.6 | 1.2±0.6 | 1.1±0.6 | 1.3±0.6 |
| Total ASF, servings/d | 1.3±0.6 | 1.5±0.6 | 1.6±0.6 | 1.8±0.5 | 2.2±0.7 | 1.7±0.7 |
| Processed meats, servings/d | 0.3±0.3 | 0.4±0.3 | 0.4±0.3 | 0.4±0.3 | 0.4±0.3 | 0.4±0.3 |
| Fish, servings/d | 0.4±0.3 | 0.3±0.2 | 0.3±0.2 | 0.3±0.2 | 0.3±0.2 | 0.3±0.2 |
| Poultry, servings/d | 0.4±0.3 | 0.3±0.3 | 0.3±0.3 | 0.3±0.2 | 0.4±0.2 | 0.3±0.3 |
| Eggs, servings/d | 0.2±0.2 | 0.2±0.2 | 0.2±0.3 | 0.2±0.2 | 0.2±0.2 | 0.2±0.2 |
| Total energy, kcal/d | 1783.4±527.1 | 1687.0±542.7 | 1858.0±752.6 | 2016.7±685.8 | 1735.6±603.8 | 1816.1±638.4 |
| TMAO-related plasma biomarkers | ||||||
| TMAO, µmol/L | 4.6 (3.1–6.9) | 4.3 (3.1–7.3) | 4.6 (3.1–7.5) | 5.1 (3.5–8.1) | 5.0 (3.4–8.2) | 4.7 (3.2–7.7) |
| Choline, µmol/L | 9.4 (8.0–10.9) | 9.6 (8.1–11.1) | 9.5 (8.0–11.1) | 9.7 (8.4–11.4) | 9.4 (8.0–11.3) | 9.5 (8.1–11.2) |
| Betaine, µmol/L | 38.0±13.9 | 37.3±12.8 | 36.8±12.4 | 37.0±13.2 | 36.0±12.5 | 37.0±13.0 |
| Carnitine, µmol/L | 35.9±8.5 | 36.8±7.7 | 37.1±8.1 | 37.7±8.3 | 38.0±8.3 | 37.1±8.2 |
| γ-butyrobetaine, µmol/L | 0.96±0.29 | 0.99±0.31 | 1.03±0.34 | 1.05±0.34 | 1.07±0.37 | 1.02±0.34 |
| Crotonobetaine, µmol/L | 0.021 (0.010–0.027) | 0.022 (0.010–0.028) | 0.023 (0.010–0.030) | 0.023 (0.010–0.029) | 0.024 (0.010–0.031) | 0.023 (0.010–0.029) |
Values are N (%), mean±SD, or median (interquartile range) at analysis baseline (1989–1990 for 3335 participants and 1996–1997 for 554 participants). Food intakes were energy-adjusted. ASF indicates animal source food; and TMAO, trimethylamine N-oxide.
*
Number of participants with joint availability of unprocessed red meat and TMAO measures was 3889.
Correlations Between Dietary Exposures and TMAO-Related Biomarkers
At baseline, small (Spearman ρ=0.05–0.07) but statistically significant positive correlations were seen between plasma TMAO levels and self-reported intakes of unprocessed red meat, total meat, fish, and total ASF but not processed meat, poultry, or eggs (Table 2. The L-carnitine metabolites γ-butyrobetaine and crotonobetaine positively correlated with intakes of unprocessed red meat, processed meat, total meat, and eggs (ρ=0.03–0.15), and inversely correlated with intakes of fish and poultry (ρ=−0.03 to −0.08). Correlations for nutrient precursors of TMAO (choline, betaine, and carnitine) are also shown. Similar modest diet-biomarker associations have been reported previously for TMAO,59–61 which could relate to imperfect measurement of self-reported diet, the temporal difference between assessment of usual dietary habits (1 year) versus shorter term dietary variations that alter TMAO levels (weeks), and interindividual biologic variation in microbial conversion of precursors to TMAO.
Associations of Dietary Exposures With the Risk of ASCVD (Main Dietary Associations)
The median follow-up was 12.5 years (range: 0.01–26.0). Numbers of events among participants included in the analysis of each dietary exposure are shown in Table 3. After adjusting for sociodemographic factors, lifestyle, dietary factors, and antibiotics use, higher intake of unprocessed red meat was associated with 15% higher incidence of ASCVD per IQR (HR=1.15 [95% CI, 1.01–1.30]; P=0.031; Table 3). Processed meat intake was associated with a similar but nonsignificant trend toward higher ASCVD risk (HR=1.11 [0.98–1.25]; P=0.089). Total meat intake (unprocessed red meat+processed meat) was associated with 22% higher incidence of ASCVD (HR=1.22 [1.07–1.39]; P=0.004).
| Unprocessed red meat | Processed meat | Total meat* | Fish | Poultry | Eggs | Total ASF† | |
|---|---|---|---|---|---|---|---|
| IQR (10th–90th), serving/d | 0.71 (0.13 to 0.84) | 0.71 (0.06 to 0.76) | 1.14 (0.25 to 1.39) | 0.46 (0.09 to 0.55) | 0.55 (0.08 to 0.63) | 0.47 (0.02 to 0.48) | 1.50 (0.93 to 2.43) |
| No. of cases/total N‡ | 1655/3889 | 1653/3891 | 1644/3871 | 1658/3891 | 1660/3898 | 1664/3908 | 1634/3843 |
| Person-years | 50 894 | 50 962 | 50 589 | 50 979 | 51 066 | 51 243 | 50 142 |
| Main dietary association§ | |||||||
| Hazard ratios (95% CI)∥ | 1.15 (1.01 to 1.30) | 1.11 (0.98 to 1.25) | 1.22 (1.07 to 1.39) | 1.00 (0.89 to 1.13) | 1.04 (0.92 to 1.18) | 1.04 (0.94 to 1.14) | 1.18 (1.03 to 1.34) |
| No. of excess events per 1000 persons per year (95% CI)¶ | 3.95 (−0.15 to 8.05) | 3.56 (−0.46 to 7.58) | 6.32 (1.77 to 10.87) | 0.49 (−3.27 to 4.25) | 1.77 (−2.13 to 5.67) | 1.43 (−2.20 to 5.06) | 5.79 (1.36 to 10.22) |
| Mediation analyses# | |||||||
| No. of excess events per 1000 persons per year (95% CI) | |||||||
| Dietary association independent of metabolites | 3.50 (−0.64 to 7.64) | 3.33 (−0.69 to 7.35) | 5.77 (1.20 to 10.34) | 0.06 (−3.76 to 3.88) | 1.80 (−2.12 to 5.72) | 1.36 (−2.27 to 4.99) | 5.22 (0.75 to 9.69) |
| Dietary association mediated via metabolites | 0.42 (0.04 to 0.85) | 0.18 (−0.05 to 0.45) | 0.49 (0.06 to 0.98) | 0.45 (0.07 to 0.88) | −0.04 (−0.26 to 0.16) | 0.06 (−0.17 to 0.29) | 0.53 (0.14 to 0.98) |
| Via γ-butyrobetaine** | 0.20 (−0.08 to 0.52) | 0.04 (−0.05 to 0.19) | 0.21 (−0.06 to 0.54) | 0.00 (−0.10 to 0.11) | 0.00 (−0.11 to 0.10) | −0.08 (−0.25 to 0.03) | 0.08 (−0.03 to 0.28) |
| Via crotonobetaine** | 0.31 (0.04 to 0.62) | 0.17 (0.01 to 0.41) | 0.39 (0.03 to 0.79) | 0.17 (0.01 to 0.39) | −0.06 (−0.24 to 0.06) | 0.09 (−0.01 to 0.26) | 0.36 (0.02 to 0.75) |
| Via TMAO** | 0.11 (−0.01 to 0.30) | 0.02 (−0.09 to 0.16) | 0.10 (−0.02 to 0.30) | 0.31 (−0.03 to 0.69) | 0.01 (−0.10 to 0.13) | 0.01 (−0.09 to 0.11) | 0.20 (−0.02 to 0.48) |
| Mediation proportions (%)†† | 10.6 (1.0 to 114.5) | 5.1 (−1.7 to 63.2) | 7.8 (1.0 to 32.7) | NA‡‡ | −2.4 (−71.3 to 20.6) | 4.2 (−30.5 to 92.7) | 9.2 (2.2 to 44.5) |
Models were adjusted for age (years), sex, race (White vs non-White), study site (4 categories), education (<high school, high school, some college, or college graduate), income (<$11 999, $12 000–24 999, $25 000–$49 999, or >$50 000), and time-varying self-reported health status (excellent, very good, good, fair, or poor), smoking status (never smoked, former smoker, or current smoker), alcohol intake (drinks/wk), physical activity (kcal/wk, log transformed for additive hazard model), antibiotic use (yes vs no), and intakes of total energy (kcal/d, log transformed for additive hazard models), fruits (servings/d), vegetables (servings/d), dietary fiber (g/d), total dairy products (servings/d), and the other animal source foods mutually adjusted (servings/d). Imputed values were used when animal source foods were adjusted covariates. ASCVD indicates atherosclerotic cardiovascular disease; ASF, animal source foods; IQR, interquintile range, comparing the midpoints of the first and fifth quintiles; and TMAO, trimethylamine N-oxide.
*
Total meat: unprocessed red meat plus processed meat.
†
Total ASF: sum of unprocessed red meat, processed meat, fish, poultry, and eggs.
‡
Given that data availability varied by dietary exposures, the number of participants included in the analyses of each exposure was different.
§
Dietary associations were estimated from models without the 3 metabolites.
∥
Hazard ratios were estimated from Cox models.
¶
No. of excess events were estimated from additive hazard models.
#Mediation analyses were performed using additive hazard models. CIs excluding zero indicate statistically significant mediated association or mediation proportion. TMAO and crotonobetaine were log transformed.
**
Refers to any paths passing through the specified metabolite. The 3 paths via γ-butyrobetaine, crotonobetaine, and TMAO were not mutually exclusive. See Figure 2 for details.
††
Mediation proportion was defined as mediated association/|(independent association + mediated association)|.
‡‡
Given that the dietary association for fish was close to null, mediation proportions for fish would not be meaningful and were not calculated.
Intakes of fish, poultry, and eggs were not significantly associated with incident ASCVD. Total ASF intake was associated with 18% higher risk (HR=1.18 [1.03–1.34]; P=0.016) per IQR. Dietary associations estimated by additive hazard models (used for mediation analyses) showed similar findings on the scale of rate difference (Table 3), although the association for unprocessed red meat did not reach statistical significance (rate difference: 4.0 events per 1000 person-years per IQR intake [95% CI, −0.1 to 8.0]; P=0.059).
Analyses of dose-response relationships between the extent of intakes of the various ASF and ASCVD (assessed using restricted cubic splines) showed key significant associations (Figure 3). In particular, increasing intakes of unprocessed red meat and total meat were dose-dependently significantly associated with increased risk of ASCVD. Processed meat ingestion trended toward both an overall and a threshold association but neither achieved statistical significance. A nonlinear relationship was suggested for poultry (P nonlinearity <0.001), with lower ASCVD risk up to a nadir of about 0.4 servings/d, and then diminished benefits thereafter; this nonlinear association was no longer statistically significant (P nonlinearity=0.083) in sensitivity analyses removing observations with extreme exposures (ie, the top and bottom 1% of the exposure distribution).
Gut Microbiota-Generated Metabolites of L-Carnitine Significantly Mediate ASF-Associated ASCVD Risk
In mediation analyses, the 3 gut microbiota-generated metabolites of dietary L-carnitine (TMAO, γ-butyrobetaine, and crotonobetaine) appeared to jointly mediate part of the association between unprocessed red meat intake and incident ASCVD. Among the total 3.92 (0.42+3.50) excess ASCVD events per 1000 person-years associated with each IQR higher intake, 0.42 events (95% CI, 0.04–0.85) or 10.6% (95% CI, 1.0–114.5) appeared attributable to plasma levels of these metabolites (Table 3). The 3 microbial metabolites also significantly mediated part of the associations of total meat and total ASF with ASCVD, accounting for 7.8% (95% CI, 1.0–32.7) and 9.2% (95% CI, 2.2–44.5) of the observed excess risk, respectively. In exploratory analyses examining path-specific mediated associations, the 4 paths via crotonobetaine alone (γ-butyrobetaine→crotonobetaine→ASCVD, γ-butyrobetaine→crotonobetaine→TMAO→ASCVD, crotonobetaine→ASCVD, and crotonobetaine→TMAO→ASCVD; see Figure 2) also significantly mediated the associations of unprocessed red meat, processed meat, total meat, and total ASF with ASCVD risk.
Interestingly, fish intake was not associated with ASCVD risk overall, but had an estimated adverse impact mediated through plasma levels of these gut microbial metabolites (0.45 excess ASCVD events per 1000 person-years [0.07–0.88] per IQR), mostly related to TMAO. No significant mediated associations were observed for any foods for the paths via γ-butyrobetaine or TMAO alone. Nor did we observe significant mediated associations by any of the nutrient precursors (ie, carnitine, choline, and betaine; Table S4).
Mediation of ASF-Associated ASCVD Risk by Traditional Risk Factors
Evaluating traditional ASCVD risk factors as mediators, neither blood cholesterol levels nor blood pressure levels significantly mediated the associations of unprocessed red meat, processed meat, or total meat with ASCVD (Table 4). In contrast, fasting blood glucose and insulin were each significant mediators of these associations; for example, mediating 26.1% (12.7–82.7) and 11.8% (4.3–43.2) of the total meat-ASCVD association, respectively. CRP also significantly mediated the associations of intakes of processed meat (13.9% [2.8–192.7]) and total meat (6.6% [0.4–27.5]), but not unprocessed red meat (0.9% [−18.6 to 21.2]), with ASCVD.
| Unprocessed red meat | Processed meat | Total meat* | |
|---|---|---|---|
| Mediator: total cholesterol | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of cholesterol | 3.9 (−0.2 to 8.0) | 3.5 (−0.5 to 7.6) | 6.3 (1.8 to 10.8) |
| Dietary association mediated via cholesterol | 0.0 (−0.1 to 0.1) | 0.0 (−0.1 to 0.2) | 0.0 (−0.1 to 0.2) |
| Mediation proportions (%) | 0.5 (−3.2 to 8.4) | 0.8 (−2.8 to 12.9) | 0.6 (−1.1 to 4.4) |
| Mediator: fasting triglycerides | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of triglycerides | 4.0 (−0.1 to 8.1) | 3.3 (−0.7 to 7.4) | 6.2 (1.6 to 10.7) |
| Dietary association mediated via triglycerides | 0.0 (−0.2 to 0.2) | 0.1 (−0.1 to 0.3) | 0.1 (−0.1 to 0.3) |
| Mediation proportions (%) | 0.3 (−10.4 to 11.3) | 1.6 (−8.5 to 30.1) | 1.6 (−2.4 to 8.8) |
| Mediator: LDL cholesterol | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of LDL cholesterol | 3.9 (−0.2 to 8.0) | 3.6 (−0.5 to 7.6) | 6.3 (1.8 to 10.8) |
| Dietary association mediated via LDL cholesterol | 0.0 (−0.1 to 0.1) | 0.0 (−0.1 to 0.1) | 0.0 (−0.1 to 0.1) |
| Mediation proportions (%) | 0.2 (−3.6 to 6.0) | 0.1 (−5.3 to 6.4) | 0.2 (−1.6 to 2.7) |
| Mediator: HDL cholesterol | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of HDL cholesterol | 3.9 (−0.2 to 8.0) | 3.6 (−0.5 to 7.6) | 6.3 (1.7 to 10.8) |
| Dietary association mediated via HDL cholesterol | 0.1 (0.0 to 0.2) | 0.0 (−0.1 to 0.1) | 0.1 (−0.1 to 0.2) |
| Mediation proportions (%) | 2.0 (−1.1 to 20.6) | −0.3 (−12.1 to 8.1) | 0.9 (−1.6 to5.7) |
| Mediator: systolic blood pressure (SBP) | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of SBP | 3.8 (−0.3 to 7.9) | 3.7 (−0.3 to 7.7) | 6.3 (1.8 to 10.9) |
| Dietary association mediated via SBP | 0.1 (−0.3 to 0.5) | −0.1 (−0.5 to 0.4) | 0.1 (−0.4 to 0.5) |
| Mediation proportions (%) | 2.1 (−21.7 to 33.7) | −1.6 (−48.3 to 18.9) | 0.8 (−10.8 to 11.2) |
| Mediator: diastolic blood pressure (DBP) | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of DBP | 3.9 (−0.2 to 8.0) | 3.7 (−0.3 to 7.6) | 6.4 (1.8 to 10.9) |
| Dietary association mediated via DBP | 0.1 (−0.1 to 0.2) | 0.0 (−0.2 to 0.2) | 0.1 (−0.1 to 0.3) |
| Mediation proportions (%) | 1.3 (−5.6 to 17.1) | 1.0 (−8.3 to 18.0) | 1.0 (−2.7 to 6.4) |
| Mediator: fasting glucose | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of glucose | 3.4 (−0.7 to 7.5) | 2.4 (−1.5 to 6.4) | 4.9 (0.4 to 9.4) |
| Dietary association mediated via glucose | 0.9 (0.4 to 1.5) | 1.2 (0.7 to 1.9) | 1.7 (1.0 to 2.6) |
| Mediation proportions (%) | 21.5 (8.5 to 175.8) | 34.0 (13.7 to 366.1) | 26.1 (12.7 to 82.7) |
| Mediator: C-reactive protein | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of C-reactive protein | 4.0 (−0.1 to 8.1) | 2.9 (−1.1 to 6.9) | 5.8 (1.3 to 10.4) |
| Dietary association mediated via C-reactive protein | 0.0 (−0.3 to 0.4) | 0.5 (0.1 to 0.9) | 0.4 (0.0 to 0.8) |
| Mediation proportions (%) | 0.9 (−18.6 to 21.2) | 13.9 (2.8 to 192.7) | 6.6 (0.4 to 27.5) |
| Mediator: fasting insulin | |||
| No. of excess events per 1000 persons per year (95% CI) | |||
| Dietary association independent of insulin | 3.7 (−0.4 to 7.8) | 3.1 (−0.9 to 7.1) | 5.7 (1.2 to 10.3) |
| Dietary association mediated via insulin | 0.4 (0.1 to 0.7) | 0.6 (0.2 to 1.0) | 0.8 (0.3 to 1.3) |
| Mediation proportions (%) | 9.6 (2.7 to 90.6) | 15.7 (4.8 to 186.6) | 11.8 (4.3 to 43.2) |
Additive hazard models were adjusted for age (years), sex, race (White vs non-White), study site (4 categories), education (<high school, high school, some college, or college graduate), income (<$11 999, $12 000–24 999, $25 000–$49 999, or >$50 000), and time-varying self-reported health status (excellent, very good, good, fair, or poor), smoking status (never smoked, former smoker, or current smoker), alcohol intake (drinks/wk), physical activity (kcal/wk, log transformed for additive hazard model), antibiotic use (yes vs no), and intakes of total energy (kcal/d, log transformed for additive hazard models), fruits (servings/d), vegetables (servings/d), dietary fiber (g/d), total dairy products (servings/d), and the other animal source foods mutually adjusted (servings/d). Imputed values were used when animal source foods were adjusted covariates. CIs excluding zero indicate statistically significant association or mediation proportion. Triglycerides, CRP, and fasting insulin were log transformed. ASCVD indicates atherosclerotic cardiovascular disease; CRP, C-reactive protein; DBP, diastolic blood pressure; HDL, high-density lipoprotein; IQR, interquintile range, comparing the mid points of the first and fifth quintiles; LDL, low-density lipoprotein; and SBP, systolic blood pressure.
*
Total meat: unprocessed red meat plus processed meat.
Sensitivity Analyses
Results for the main dietary associations and mediation by the TMAO-related metabolites were not appreciably changed when using simple updating of dietary intakes in place of cumulative updating (Table S5). Results were also similar after further adjustments for additional CVD risk factors (Table S6). Spearman correlations between eGFR and TMAO, γ-butyrobetaine, and crotonobetaine were −0.31, −0.34, and −0.38, respectively (P<0.001 each). After additional adjustment for eGFR which could be both a confounder and mediator for the metabolites-ASCVD associations, although the magnitude of main dietary associations remained similar, mediation proportions were attenuated and no longer statistically significant. Examination of the modeling results revealed that this was related to attenuation of the mediator-outcome association (ie, associations of these metabolites with ASCVD), rather than the exposure-mediator association. Renal function did not significantly modify the associations between any of the ASF and the risk of ASCVD (P interaction >0.10 for each). Exploratory analyses identified no significant interactions of the ASF-ASCVD relationships by age, sex, race and ethnicity, education level, or smoking status (P interaction >0.0014 each).
Discussion
In this large, community-based prospective cohort of older adults, higher intakes of unprocessed red meat, total meat, and total ASF were each associated with higher risk of ASCVD, with processed meats trending toward higher risk. These associations were partly (≈8%–11%) mediated by plasma levels of 3 dietary L-carnitine-derived gut microbiota-generated metabolites: TMAO, γ-butyrobetaine, and crotonobetaine. Path-specific analyses suggested that plasma crotonobetaine accounted for the largest proportion of the observed mediation. Intakes of fish, poultry, and eggs were not statistically significantly associated with ASCVD. To our knowledge, this is the first study to investigate the association of ASF with ASCVD and potential mediation by gut microbiota-generated TMAO-related metabolites.
Prior studies of ASF and CVD have primarily included middle-aged participants. In these studies, processed meat intake most consistently associates with higher risk, while associations for unprocessed red meat have been smaller and less consistent.29,30,62–72 A recent meta-analysis suggests similar overall magnitudes of associations as in our investigation, with HRs (95% CIs) for CVD of 1.18 (1.10–1.30) for processed meat and 1.08 (1.03–1.16) for unprocessed red meat (scaled to the same servings as in our study).30 Associations for fish intake have generally been specific to coronary events, especially sudden death,28,72–74 but less consistently with stroke or total ASCVD30—for example, a recent meta-analysis identified a pooled HR for CVD per 2 weekly fish servings of 1.00 (95% CI, 0.98–1.02).30 Eggs and poultry have also generally had minimal or neutral associations with ASCVD in prior analyses.30,71,72,75–78 Our findings for ASF and ASCVD in this population of older US adults, average age 73 years at baseline and followed for an average of 13 years, were generally consistent with these previous studies. We demonstrated a linear dose-response relationship between higher unprocessed red meat and total meat intake and higher incidence of ASCVD later in life. Processed meat was associated with a similar magnitude of increased risk, although the association did not achieve statistical significance, perhaps related to the relatively low intake in CHS of processed meat (median: 0.2 servings/d) versus unprocessed red meat (median: 0.4 servings/d).
Several ingredients and mechanisms have been proposed to explain potential harmful effects of meat intake on ASCVD. These include contents of saturated fat, cholesterol, and heme iron in red meats, as well as sodium, nitrites, and high temperature cooking of processed meats.66 However, true mechanisms are surprisingly poorly understood. Mounting evidence indicates heterogeneous health effects of saturated fat on blood cholesterol levels and ASCVD depending on the type of saturated fat as well as the food source.4,79 A consensus has also emerged that dietary cholesterol has little meaningful effects on blood cholesterol levels or ASCVD risk at amounts commonly consumed.80,81 Consistent with this, in our analysis, neither blood cholesterol levels nor blood pressure levels significantly mediated the associations between unprocessed red meat, processed meat, or total meat and incidence of ASCVD. In contrast, blood glucose levels and insulin sensitivity (measured by fasting insulin) mediated a significant proportion of the meat-ASCVD associations. Red meat is the major source of dietary heme iron, which is implicated as a causal factor in development of type 2 diabetes82–85 and associated with increased CVD.86 In a previous mediation analysis, increased cardiovascular mortality associated with processed red meat intake was mediated by both heme iron (24.1%) and nitrite (72.0%) intake; and with unprocessed red meat intake, mediated by heme iron (20.8%) with a large portion of the remaining excess risk unexplained.70 Our findings support mechanisms related to glucose-insulin homeostasis, and therefore potentially heme content, as one important pathway whereby meat consumption may influence ASCVD. TMAO-related metabolites may play a role here, since TMAO has been mechanistically linked to hyperglycemia and insulin resistance.87,88 Systematic inflammation as assessed by CRP also mediated a significant proportion of the association between processed meat, but not unprocessed red meat, and ASCVD.
Our novel findings further suggest that L-carnitine derived microbiome metabolites play a larger mediating role in meat-ASCVD associations than blood pressure or blood cholesterol levels. This result is consistent with, and may partly help explain, the neutral associations of saturated fat consumption with CVD4,89 and suggest that attention to other meat constituents and risk pathways is needed. The interplay between diet, the gut microbiota, and microbial-generated metabolites increasingly appears to be a novel pathway linking ASF, especially red meat, to cardiovascular health.5,90 Dietary L-carnitine, a nutrient abundant almost exclusively in red meat, can be metabolized to γ-butyrobetaine and crotonobetaine, and ultimately TMAO through the action of gut microbiota and hepatic flavin monooxygenases (Figure 1)16,27; and habitual red meat ingestion increases plasma TMAO more than other animal or plant-based protein sources.7 Our findings suggest that TMAO, γ-butyrobetaine, and crotonobetaine together explain about 8% to 11% of observed excess ASCVD risk associated with intakes of unprocessed red meat and total meat. Exploratory analyses suggested that γ-butyrobetaine and especially crotonobetaine may be at least as important as TMAO in such mediation, important given that these specific metabolites are derived from the carnitine-pathway, rather than the alternative choline-pathway, for TMAO production.
In experimental studies, TMAO promotes cholesterol accumulation in macrophages by upregulating cell surface expression of the proatherogenic scavenger receptors CD36 and SR-A1 (class A1 scavenger receptors)8; inhibits reverse cholesterol transport and alters sterol metabolism16; enhances vascular inflammation through activation of mitogen-activated protein kinase and nuclear factor-κB signaling9 and inflammasome activation91; impairs endothelial function by increasing superoxide-associated oxidative stress13; and promotes platelet hyperresponsiveness and thrombosis potential by enhancing stimulus-dependent Ca2+ release from intracellular stores.14 TMAO also induces hyperglycemia by binding to protein kinase R-like endoplasmic reticulum stress kinase PERK (EIF2AK3).88 Possible proatherogenic effects of γ-butyrobetaine and crotonobetaine have not been reported. Our findings highlight the need to investigate whether these metabolites have independent physiological effects or simply provide an additional or even superior measure of overall tissue exposure to TMAO.
The estimated mediation proportions were attenuated by about half following adjustment for eGFR. Because these metabolites are renally cleared, impaired renal function could confound their associations with ASCVD. However, mechanistic studies demonstrate that TMAO directly causes renal tubulointerstitial fibrosis, reduced renal filtration, and elevated cystatin-C levels.53 In addition, suppression of TMAO generation prevents renal impairment in animal models.54 Thus, impaired renal function may also be an intermediate outcome (ie, mediator) on the causal pathway between these TMAO-related metabolites and ASCVD. Future experimental studies are needed to further investigate the potential interplay and extent of confounding versus mediation between intakes of ASF, these gut microbial metabolites, renal function, and ASCVD.
Prior work by our group and others has demonstrated that cardiovascular benefits of fish consumption or omega-3 supplementation may depend on outcomes examined, with stronger associations for CHD, especially fatal CHD or arrhythmic cardiac death, than stroke or total ASCVD30,74,92; and on fish preparation methods, with protective associations for tuna fish or broiled/baked fish but not fried fish or fish sandwiches.74 The specificity for coronary events is consistent with the experimental impact of omega-3 fatty acids on stabilization of partially depolarized, acutely ischemic myocytes, reducing susceptibility to acute ventricular arrhythmias.93 Our focus in the present analysis was on total fish consumption and total ASCVD, and the absence of significant association with this compositive end point is consistent with extant literature.30,92 Nonetheless, while there was no evidence of overall association (HR=1.00), our mediation analyses suggest the relationship between fish intake and these TMAO-related plasma metabolites was associated with some excess risk. Other beneficial compounds in fish could offset this estimated harm, so the overall association of fish with total ASCVD was neutral. In contrast to meats, the largest mediated association for fish was via TMAO, consistent with fish being a rich source of choline (a precursor of TMAO) but not L-carnitine (a precursor of γ-butyrobetaine, crotonobetaine, and TMAO).
The association of poultry consumption with CVD has not been well-studied,30 especially in older populations. Our exploratory analyses suggested a possible nonlinear dose-response, with lowest risk at about 0.4 servings/d. Adequate intakes of protein can help prevent aging-related loss of muscle mass, improving physical functioning and long-term health outcomes later in life.1–3 Our findings suggest that poultry might be a healthy source of protein for older adults when consumed moderately. However, the nonlinear analyses were exploratory, so conclusions based on them may be due to chance, and should be interpreted with caution until confirmed in other investigations.
Our study has several strengths. The relationship between ASF, plasma levels of microbiome-derived TMAO-related metabolites, and ASCVD events has not been fully established and warrants careful investigation, especially in community-based prospective cohorts such as the CHS with well-measured metabolite biomarkers, ASCVD risk factors (including detailed sociodemographics, traditional CVD risk factors, and lifestyle habits), and a sufficient number of outcomes. Such findings are less subject to bias, residual confounding, and reverse causation; have greater statistical power; and have greater generalizability than many of the prior studies. We evaluated multiple ASF in relation to risk of ASCVD in older adults, the age group at the highest risk; and further investigated potential mediation by a novel set of microbiome-derived metabolites, providing an important new piece of evidence in the puzzle of diet, the microbiome, and ASCVD.
Potential limitations should be considered. The observational design cannot exclude residual confounding. However, we adjusted for a broad range of well-measured risk factors for ASCVD, and results were robust except for additional adjustment for eGFR, which could be both a mediator and confounder for the metabolites-ASCVD associations. Factors known to influence TMAO’s generation such as host hepatic flavin-containing monooxygenase and choline TMA-lyase carried by some bacterial species in the gut were not measured; yet, there is no established evidence that these enzymes alter dietary intake or affect CVD risk via paths independent of TMAO. Thus, the potential for major confounding that could fully account for the observed associations, conditional upon all other covariates in the model, is not high. Although we applied statistical methods for causal mediation analysis, the study findings are observational and cannot prove causality. Dietary habits were self-reported, which could cause nondifferential measurement errors with respective to ASCVD and metabolite measurements. Because dietary data in CHS were validated against nutrients rather than foods, we could not perform correction for such measurement errors, although we took advantage of serial measures from 2 validated FFQs. Analyses for secondary dietary exposures should be interpreted with caution, although the total meat-ASCVD association (P=0.004) would remain significant after adjusting for multiple comparisons (adjusted α=0.05/4=0.0125). Our findings may not be generalizable to young populations, different races, or other nations.
Conclusions
In this large, community-based cohort of older US adults, higher intakes of unprocessed red meat, total meat, and total ASF were associated with higher incidence of ASCVD, partly explained by plasma levels of γ-butyrobetaine, crotonobetaine, and TMAO. The higher risk of ASCVD associated with meats further appeared partly mediated by glucose-insulin homeostasis and systematic inflammation but not blood pressure or blood cholesterol levels. These novel findings support a biochemical link between dietary meat intake, carnitine-related gut microbiome pathways, and ASCVD.
Article Information
Supplemental Material
Expanded Materials & Methods
Figures S1–S3
Tables S1–S6
Acknowledgments
Sincere thanks to all CHS (Cardiovascular Health Study) participants, CHS investigators, and institutions (see www.chs-nhlbi.org).
Footnote
Nonstandard Abbreviations and Acronyms
- ASCVD
- atherosclerotic cardiovascular disease
- ASF
- animal source food
- CHD
- coronary heart disease
- CHS
- Cardiovascular Health Study
- CRP
- C-reactive protein
- eGFR
- estimated glomerular filtration rate
- FFQ
- food frequency questionnaire
- HDL
- high-density lipoprotein
- HR
- hazard ratio
- IQR
- interquintile range
- LDL
- low-density lipoprotein
- TMAO
- trimethylamine N-oxide
Supplemental Material
References
1.
Hackney KJ, Trautman K, Johnson N, Mcgrath R, Stastny S. Protein and muscle health during aging: benefits and concerns related to animal-based protein. Anim Front. 2019;9:12–17. doi: 10.1093/af/vfz030
2.
Wolfe RR. The role of dietary protein in optimizing muscle mass, function and health outcomes in older individuals. Br J Nutr. 2012;108 Suppl 2:S88–S93. doi: 10.1017/S0007114512002590
3.
Wolfe RR, Miller SL, Miller KB. Optimal protein intake in the elderly. Clin Nutr. 2008;27:675–684. doi: 10.1016/j.clnu.2008.06.008
4.
Astrup A, Magkos F, Bier DM, Brenna JT, de Oliveira Otto MC, Hill JO, King JC, Mente A, Ordovas JM, Volek JS, et al. Saturated fats and health: a reassessment and proposal for food-based recommendations: JACC state-of-the-art review. J Am Coll Cardiol. 2020;76:844–857. doi: 10.1016/j.jacc.2020.05.077
5.
Tang WHW, Bäckhed F, Landmesser U, Hazen SL. Intestinal microbiota in cardiovascular health and disease: JACC state-of-the-art review. J Am Coll Cardiol. 2019;73:2089–2105. doi: 10.1016/j.jacc.2019.03.024
6.
Koeth RA, Levison BS, Culley MK, Buffa JA, Wang Z, Gregory JC, Org E, Wu Y, Li L, Smith JD, et al. γ-Butyrobetaine is a proatherogenic intermediate in gut microbial metabolism of L-carnitine to TMAO. Cell Metab. 2014;20:799–812. doi: 10.1016/j.cmet.2014.10.006
7.
Wang Z, Bergeron N, Levison BS, Li XS, Chiu S, Jia X, Koeth RA, Li L, Wu Y, Tang WHW, et al. Impact of chronic dietary red meat, white meat, or non-meat protein on trimethylamine N-oxide metabolism and renal excretion in healthy men and women. Eur Heart J. 2019;40:583–594. doi: 10.1093/eurheartj/ehy799
8.
Wang Z, Klipfell E, Bennett BJ, Koeth R, Levison BS, Dugar B, Feldstein AE, Britt EB, Fu X, Chung YM, et al. Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature. 2011;472:57–63. doi: 10.1038/nature09922
9.
Seldin MM, Meng Y, Qi H, Zhu W, Wang Z, Hazen SL, Lusis AJ, Shih DM. Trimethylamine N-oxide promotes vascular inflammation through signaling of mitogen-activated protein kinase and nuclear factor-κB. J Am Heart Assoc. 2016;5:e002767. doi: 10.1161/JAHA.115.002767
10.
Chen ML, Zhu XH, Ran L, Lang HD, Yi L, Mi MT. Trimethylamine-N-oxide induces vascular inflammation by activating the NLRP3 inflammasome through the SIRT3-SOD2-mtROS signaling pathway. J Am Heart Assoc. 2017;6:e006347. doi: 10.1161/JAHA.117.006347
11.
Yue C, Yang X, Li J, Chen X, Zhao X, Chen Y, Wen Y. Trimethylamine N-oxide prime NLRP3 inflammasome via inhibiting ATG16L1-induced autophagy in colonic epithelial cells. Biochem Biophys Res Commun. 2017;490:541–551. doi: 10.1016/j.bbrc.2017.06.075
12.
Boini KM, Hussain T, Li PL, Koka S. Trimethylamine-N-Oxide instigates NLRP3 inflammasome activation and endothelial dysfunction. Cell Physiol Biochem. 2017;44:152–162. doi: 10.1159/000484623
13.
Brunt VE, Gioscia-Ryan RA, Casso AG, VanDongen NS, Ziemba BP, Sapinsley ZJ, Richey JJ, Zigler MC, Neilson AP, Davy KP, et al. Trimethylamine-N-Oxide promotes age-related vascular oxidative stress and endothelial dysfunction in mice and healthy humans. Hypertension. 2020;76:101–112. doi: 10.1161/HYPERTENSIONAHA.120.14759
14.
Zhu W, Gregory JC, Org E, Buffa JA, Gupta N, Wang Z, Li L, Fu X, Wu Y, Mehrabian M, et al. Gut microbial metabolite TMAO enhances platelet hyperreactivity and thrombosis risk. Cell. 2016;165:111–124. doi: 10.1016/j.cell.2016.02.011
15.
Zhu W, Wang Z, Tang WHW, Hazen SL. Gut microbe-generated trimethylamine N-Oxide from dietary choline is prothrombotic in subjects. Circulation. 2017;135:1671–1673. doi: 10.1161/CIRCULATIONAHA.116.025338
16.
Koeth RA, Wang Z, Levison BS, Buffa JA, Org E, Sheehy BT, Britt EB, Fu X, Wu Y, Li L, et al. Intestinal microbiota metabolism of L-carnitine, a nutrient in red meat, promotes atherosclerosis. Nat Med. 2013;19:576–585. doi: 10.1038/nm.3145
17.
Mueller DM, Allenspach M, Othman A, Saely CH, Muendlein A, Vonbank A, Drexel H, von Eckardstein A. Plasma levels of trimethylamine-N-oxide are confounded by impaired kidney function and poor metabolic control. Atherosclerosis. 2015;243:638–644. doi: 10.1016/j.atherosclerosis.2015.10.091
18.
Yin J, Liao SX, He Y, Wang S, Xia GH, Liu FT, Zhu JJ, You C, Chen Q, Zhou L, et al. Dysbiosis of gut microbiota with reduced trimethylamine-N-Oxide level in patients with large-artery atherosclerotic stroke or transient ischemic attack. J Am Heart Assoc. 2015;4:e002699. doi: 10.1161/JAHA.115.002699
19.
Tang WH, Wang Z, Levison BS, Koeth RA, Britt EB, Fu X, Wu Y, Hazen SL. Intestinal microbial metabolism of phosphatidylcholine and cardiovascular risk. N Engl J Med. 2013;368:1575–1584. doi: 10.1056/NEJMoa1109400
20.
Farhangi MA. Gut microbiota-dependent trimethylamine N-oxide and all-cause mortality: findings from an updated systematic review and meta-analysis. Nutrition. 2020;78:110856. doi: 10.1016/j.nut.2020.110856
21.
Qi J, You T, Li J, Pan T, Xiang L, Han Y, Zhu L. Circulating trimethylamine N-oxide and the risk of cardiovascular diseases: a systematic review and meta-analysis of 11 prospective cohort studies. J Cell Mol Med. 2018;22:185–194. doi: 10.1111/jcmm.13307
22.
Schiattarella GG, Sannino A, Toscano E, Giugliano G, Gargiulo G, Franzone A, Trimarco B, Esposito G, Perrino C. Gut microbe-generated metabolite trimethylamine-N-oxide as cardiovascular risk biomarker: a systematic review and dose-response meta-analysis. Eur Heart J. 2017;38:2948–2956. doi: 10.1093/eurheartj/ehx342
23.
Heianza Y, Ma W, Manson JE, Rexrode KM, Qi L. Gut microbiota metabolites and risk of major adverse cardiovascular disease events and death: a systematic review and meta-analysis of prospective studies. J Am Heart Assoc. 2017;6:e004947. doi: 10.1161/JAHA.116.004947
24.
Abbasi J. TMAO and heart disease: the new red meat risk? JAMA. 2019;321:2149–2151. doi: 10.1001/jama.2019.3910
25.
Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A. The cardiovascular health study: design and rationale. Ann Epidemiol. 1991;1:263–276. doi: 10.1016/1047-2797(91)90005-w
26.
Tell GS, Fried LP, Hermanson B, Manolio TA, Newman AB, Borhani NO. Recruitment of adults 65 years and older as participants in the cardiovascular health study. Ann Epidemiol. 1993;3:358–366. doi: 10.1016/1047-2797(93)90062-9
27.
Koeth RA, Lam-Galvez BR, Kirsop J, Wang Z, Levison BS, Gu X, Copeland MF, Bartlett D, Cody DB, Dai HJ, et al. l-Carnitine in omnivorous diets induces an atherogenic gut microbial pathway in humans. J Clin Invest. 2019;129:373–387. doi: 10.1172/JCI94601
28.
Leung Yinko SS, Stark KD, Thanassoulis G, Pilote L. Fish consumption and acute coronary syndrome: a meta-analysis. Am J Med. 2014;127:848–57.e2. doi: 10.1016/j.amjmed.2014.04.016
29.
Micha R, Wallace SK, Mozaffarian D. Red and processed meat consumption and risk of incident coronary heart disease, stroke, and diabetes mellitus: a systematic review and meta-analysis. Circulation. 2010;121:2271–2283. doi: 10.1161/CIRCULATIONAHA.109.924977
30.
Zhong VW, Van Horn L, Greenland P, Carnethon MR, Ning H, Wilkins JT, Lloyd-Jones DM, Allen NB. Associations of processed meat, unprocessed red meat, poultry, or fish intake with incident cardiovascular disease and all-cause mortality. JAMA Intern Med. 2020;180:503–512. doi: 10.1001/jamainternmed.2019.6969
31.
Micha R, Shulkin ML, Peñalvo JL, Khatibzadeh S, Singh GM, Rao M, Fahimi S, Powles J, Mozaffarian D. Etiologic effects and optimal intakes of foods and nutrients for risk of cardiovascular diseases and diabetes: systematic reviews and meta-analyses from the nutrition and chronic diseases expert group (NutriCoDE). PLoS One. 2017;12:e0175149. doi: 10.1371/journal.pone.0175149
32.
Chowdhury R, Stevens S, Gorman D, Pan A, Warnakula S, Chowdhury S, Ward H, Johnson L, Crowe F, Hu FB, et al. Association between fish consumption, long chain omega 3 fatty acids, and risk of cerebrovascular disease: systematic review and meta-analysis. BMJ. 2012;345:e6698. doi: 10.1136/bmj.e6698
33.
Schwingshackl L, Schwedhelm C, Hoffmann G, Lampousi AM, Knüppel S, Iqbal K, Bechthold A, Schlesinger S, Boeing H. Food groups and risk of all-cause mortality: a systematic review and meta-analysis of prospective studies. Am J Clin Nutr. 2017;105:1462–1473. doi: 10.3945/ajcn.117.153148
34.
Kumanyika S, Tell GS, Shemanski L, Polak J, Savage PJ. Eating patterns of community-dwelling older adults: the cardiovascular health study. Ann Epidemiol. 1994;4:404–415. doi: 10.1016/1047-2797(94)90076-0
35.
Kumanyika SK, Tell GS, Shemanski L, Martel J, Chinchilli VM. Dietary assessment using a picture-sort approach. Am J Clin Nutr. 1997;65(4 Suppl):1123S–1129S. doi: 10.1093/ajcn/65.4.1123S
36.
Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, Hennekens CH, Speizer FE. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. 1985;122:51–65. doi: 10.1093/oxfordjournals.aje.a114086
37.
Feskanich D, Rimm EB, Giovannucci EL, Colditz GA, Stampfer MJ, Litin LB, Willett WC. Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. J Am Diet Assoc. 1993;93:790–796. doi: 10.1016/0002-8223(93)91754-e
38.
Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett WC. Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol. 1992;135:1114–26; discussion 1127. doi: 10.1093/oxfordjournals.aje.a116211
39.
Hu FB, Rimm E, Smith-Warner SA, Feskanich D, Stampfer MJ, Ascherio A, Sampson L, Willett WC. Reproducibility and validity of dietary patterns assessed with a food-frequency questionnaire. Am J Clin Nutr. 1999;69:243–249. doi: 10.1093/ajcn/69.2.243
40.
Mozaffarian D, Kumanyika SK, Lemaitre RN, Olson JL, Burke GL, Siscovick DS. Cereal, fruit, and vegetable fiber intake and the risk of cardiovascular disease in elderly individuals. JAMA. 2003;289:1659–1666. doi: 10.1001/jama.289.13.1659
41.
Willett, W. Nutritional epidemiology. New York: Oxford University Press; 2013.
42.
Geffken DF, Cushman M, Burke GL, Polak JF, Sakkinen PA, Tracy RP. Association between physical activity and markers of inflammation in a healthy elderly population. Am J Epidemiol. 2001;153:242–250. doi: 10.1093/aje/153.3.242
43.
Taylor HL, Jacobs DR, Schucker B, Knudsen J, Leon AS, Debacker G. A questionnaire for the assessment of leisure time physical activities. J Chronic Dis. 1978;31:741–755. doi: 10.1016/0021-9681(78)90058-9
44.
Hussein AA, Gottdiener JS, Bartz TM, Sotoodehnia N, DeFilippi C, See V, Deo R, Siscovick D, Stein PK, Lloyd-Jones D. Inflammation and sudden cardiac death in a community-based population of older adults: the cardiovascular health study. Heart Rhythm. 2013;10:1425–1432. doi: 10.1016/j.hrthm.2013.07.004
45.
Steubl D, Buzkova P, Garimella PS, Ix JH, Devarajan P, Bennett MR, Chaves PHM, Shlipak MG, Bansal N, Sarnak MJ. Association of serum uromodulin with ESKD and kidney function decline in the elderly: the cardiovascular health study. Am J Kidney Dis. 2019;74:501–509. doi: 10.1053/j.ajkd.2019.02.024
46.
Weiner DE, Tighiouart H, Amin MG, Stark PC, MacLeod B, Griffith JL, Salem DN, Levey AS, Sarnak MJ. Chronic kidney disease as a risk factor for cardiovascular disease and all-cause mortality: a pooled analysis of community-based studies. J Am Soc Nephrol. 2004;15:1307–1315. doi: 10.1097/01.asn.0000123691.46138.e2
47.
Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI, Greene T, Kusek JW, Manzi J, Van Lente F, Zhang YL, et al;. CKD-EPI Investigators. Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med. 2012;367:20–29. doi: 10.1056/NEJMoa1114248
48.
American Diabetes Association; 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes—2020. Diabetes Care 1. January 2020;43(Supplement_1):S14–S31. https://doi.org/10.2337/dc20-S002
49.
Ives DG, Fitzpatrick AL, Bild DE, Psaty BM, Kuller LH, Crowley PM, Cruise RG, Theroux S. Surveillance and ascertainment of cardiovascular events. the cardiovascular health study. Ann Epidemiol. 1995;5:278–285. doi: 10.1016/1047-2797(94)00093-9
50.
Price TR, Psaty B, O’Leary D, Burke G, Gardin J. Assessment of cerebrovascular disease in the Cardiovascular Health Study. Ann Epidemiol. 1993;3:504–507. doi: 10.1016/1047-2797(93)90105-d
52.
Lee Y, Nemet I, Wang Z, Lai HTM, de Oliveira Otto MC, Lemaitre RN, Fretts AM, Sotoodehnia N, Budoff M, DiDonato JA, et al. Longitudinal plasma measures of trimethylamine N-Oxide and risk of atherosclerotic cardiovascular disease events in community-based older adults. J Am Heart Assoc. 2021;10:e020646. doi: 10.1161/JAHA.120.020646
53.
Tang WH, Wang Z, Kennedy DJ, Wu Y, Buffa JA, Agatisa-Boyle B, Li XS, Levison BS, Hazen SL. Gut microbiota-dependent trimethylamine N-oxide (TMAO) pathway contributes to both development of renal insufficiency and mortality risk in chronic kidney disease. Circ Res. 2015;116:448–455. doi: 10.1161/CIRCRESAHA.116.305360
54.
Gupta N, Buffa JA, Roberts AB, Sangwan N, Skye SM, Li L, Ho KJ, Varga J, DiDonato JA, Tang WHW, et al. Targeted inhibition of gut microbial trimethylamine N-Oxide production reduces renal tubulointerstitial fibrosis and functional impairment in a murine model of chronic kidney disease. Arterioscler Thromb Vasc Biol. 2020;40:1239–1255. doi: 10.1161/ATVBAHA.120.314139
55.
Arnold AM, Kronmal RA. Multiple imputation of baseline data in the cardiovascular health study. Am J Epidemiol. 2003;157:74–84. doi: 10.1093/aje/kwf156
56.
Huang YT, Yang HI. Causal mediation analysis of survival outcome with multiple mediators. Epidemiology. 2017;28:370–378. doi: 10.1097/EDE.0000000000000651
57.
Lange T, Hansen JV. Direct and indirect effects in a survival context. Epidemiology. 2011;22:575–581. doi: 10.1097/EDE.0b013e31821c680c
58.
VanderWeele TJ. Mediation analysis: a practitioner’s guide. Annu Rev Public Health. 2016;37:17–32. doi: 10.1146/annurev-publhealth-032315-021402
59.
Hamaya R, Ivey KL, Lee DH, Wang M, Li J, Franke A, Sun Q, Rimm EB. Association of diet with circulating trimethylamine-N-oxide concentration. Am J Clin Nutr. 2020;112:1448–1455. doi: 10.1093/ajcn/nqaa225
60.
Krüger R, Merz B, Rist MJ, Ferrario PG, Bub A, Kulling SE, Watzl B. Associations of current diet with plasma and urine TMAO in the karmen study: direct and indirect contributions. Mol Nutr Food Res. 2017;61:1700363. doi: 10.1002/mnfr.201700363
61.
Yang JJ, Shu XO, Herrington DM, Moore SC, Meyer KA, Ose J, Menni C, Palmer ND, Eliassen H, Harada S, et al. Circulating trimethylamine N-oxide in association with diet and cardiometabolic biomarkers: an international pooled analysis. Am J Clin Nutr. 2021;113:1145–1156. doi: 10.1093/ajcn/nqaa430
62.
Zeraatkar D, Han MA, Guyatt GH, Vernooij RWM, El Dib R, Cheung K, Milio K, Zworth M, Bartoszko JJ, Valli C, et al. Red and processed meat consumption and risk for all-cause mortality and cardiometabolic outcomes: a systematic review and meta-analysis of cohort studies. Ann Intern Med. 2019;171:703–710. doi: 10.7326/M19-0655
63.
Cui K, Liu Y, Zhu L, Mei X, Jin P, Luo Y. Association between intake of red and processed meat and the risk of heart failure: a meta-analysis. BMC Public Health. 2019;19:354. doi: 10.1186/s12889-019-6653-0
64.
Abete I, Romaguera D, Vieira AR, Lopez de Munain A, Norat T. Association between total, processed, red and white meat consumption and all-cause, CVD and IHD mortality: a meta-analysis of cohort studies. Br J Nutr. 2014;112:762–775. doi: 10.1017/S000711451400124X
65.
Guasch-Ferré M, Satija A, Blondin SA, Janiszewski M, Emlen E, O’Connor LE, Campbell WW, Hu FB, Willett WC, Stampfer MJ. Meta-analysis of randomized controlled trials of red meat consumption in comparison with various comparison diets on cardiovascular risk factors. Circulation. 2019;139:1828–1845. doi: 10.1161/CIRCULATIONAHA.118.035225
66.
Micha R, Michas G, Mozaffarian D. Unprocessed red and processed meats and risk of coronary artery disease and type 2 diabetes–an updated review of the evidence. Curr Atheroscler Rep. 2012;14:515–524. doi: 10.1007/s11883-012-0282-8
67.
Yang C, Pan L, Sun C, Xi Y, Wang L, Li D. Red meat consumption and the risk of stroke: a dose–response meta-analysis of prospective cohort studies. Journal of Stroke and Cerebrovascular Diseases. 2016;25:1177–1186. doi: 10.1161/CIRCULATIONAHA.109.924977
68.
Kim K, Hyeon J, Lee SA, Kwon SO, Lee H, Keum N, Lee JK, Park SM. Role of total, red, processed, and white meat consumption in stroke incidence and mortality: a systematic review and meta-analysis of prospective cohort studies. J Am Heart Assoc. 2017;6:e005983. doi: 10.1161/JAHA.117.005983
69.
Micha R, Peñalvo JL, Cudhea F, Imamura F, Rehm CD, Mozaffarian D. Association between dietary factors and mortality from heart disease, stroke, and type 2 diabetes in the united states. JAMA. 2017;317:912–924. doi: 10.1001/jama.2017.0947
70.
Etemadi A, Sinha R, Ward MH, Graubard BI, Inoue-Choi M, Dawsey SM, Abnet CC. Mortality from different causes associated with meat, heme iron, nitrates, and nitrites in the NIH-AARP Diet and Health Study: population based cohort study. BMJ. 2017;357:j1957. doi: 10.1136/bmj.j1957
71.
Key TJ, Appleby PN, Bradbury KE, Sweeting M, Wood A, Johansson I, Kühn T, Steur M, Weiderpass E, Wennberg M, et al. Consumption of meat, fish, dairy products, and eggs and risk of ischemic heart disease. Circulation. 2019;139:2835–2845. doi: 10.1161/CIRCULATIONAHA.118.038813
72.
Bernstein AM, Sun Q, Hu FB, Stampfer MJ, Manson JE, Willett WC. Major dietary protein sources and risk of coronary heart disease in women. Circulation. 2010;122:876–883. doi: 10.1161/CIRCULATIONAHA.109.915165
73.
Djoussé L, Akinkuolie AO, Wu JH, Ding EL, Gaziano JM. Fish consumption, omega-3 fatty acids and risk of heart failure: a meta-analysis. Clin Nutr. 2012;31:846–853. doi: 10.1016/j.clnu.2012.05.010
74.
Mozaffarian D, Lemaitre RN, Kuller LH, Burke GL, Tracy RP, Siscovick DS; Cardiovascular Health Study. Cardiac benefits of fish consumption may depend on the type of fish meal consumed: the Cardiovascular Health Study. Circulation. 2003;107:1372–1377. doi: 10.1161/01.cir.0000055315.79177.16
75.
Drouin-Chartier JP, Chen S, Li Y, Schwab AL, Stampfer MJ, Sacks FM, Rosner B, Willett WC, Hu FB, Bhupathiraju SN. Egg consumption and risk of cardiovascular disease: three large prospective US cohort studies, systematic review, and updated meta-analysis. BMJ. 2020;368:m513. doi: 10.1136/bmj.m513
76.
Dehghan M, Mente A, Yusuf S. Eggs and diabetes: 1 daily egg a safe bet? Am J Clin Nutr. 2020;112:503–504. doi: 10.1093/ajcn/nqaa183
77.
Qin C, Lv J, Guo Y, Bian Z, Si J, Yang L, Chen Y, Zhou Y, Zhang H, Liu J, et al;. China Kadoorie Biobank Collaborative Group. Associations of egg consumption with cardiovascular disease in a cohort study of 0.5 million Chinese adults. Heart. 2018;104:1756–1763. doi: 10.1136/heartjnl-2017-312651
78.
Zhong VW, Van Horn L, Cornelis MC, Wilkins JT, Ning H, Carnethon MR, Greenland P, Mentz RJ, Tucker KL, Zhao L, et al. Associations of dietary cholesterol or egg consumption with incident cardiovascular disease and mortality. JAMA. 2019;321:1081–1095. doi: 10.1001/jama.2019.1572
79.
Astrup A, Bertram HC, Bonjour JP, de Groot LC, de Oliveira Otto MC, Feeney EL, Garg ML, Givens I, Kok FJ, Krauss RM, et al. WHO draft guidelines on dietary saturated and trans fatty acids: time for a new approach? BMJ. 2019;366:l4137. doi: 10.1136/bmj.l4137
80.
Carson JAS, Lichtenstein AH, Anderson CAM, Appel LJ, Kris-Etherton PM, Meyer KA, Petersen K, Polonsky T, Van Horn L; American Heart Association Nutrition Committee of the Council on Lifestyle and Cardiometabolic Health; Council on Arteriosclerosis, Thrombosis and Vascular Biology; Council on Cardiovascular and Stroke Nursing; Council on Clinical Cardiology; Council on Peripheral Vascular Disease; and Stroke Council. Dietary cholesterol and cardiovascular risk: a science advisory from the american heart association. Circulation. 2020;141:e39–e53. doi: 10.1161/CIR.0000000000000743
81.
2015 DGAC. Scientific report of the 2015 dietary guidelines advisory committee: advisory report to the secretary of health and human services and the secretary of agriculture.
82.
Creighton Mitchell T, McClain DA. Diabetes and hemochromatosis. Curr Diab Rep. 2014;14:488. doi: 10.1007/s11892-014-0488-y
83.
Wang X, Fang X, Wang F. Pleiotropic actions of iron balance in diabetes mellitus. Rev Endocr Metab Disord. 2015;16:15–23. doi: 10.1007/s11154-014-9303-y
84.
Zhao L, Lian J, Tian J, Shen Y, Ping Z, Fang X, Min J, Wang F. Dietary intake of heme iron and body iron status are associated with the risk of gestational diabetes mellitus: a systematic review and meta-analysis. Asia Pac J Clin Nutr. 2017;26:1092–1106. doi: 10.6133/apjcn.022017.09
85.
Gáll T, Balla G, Balla J. Heme, heme oxygenase, and endoplasmic reticulum stress-a new insight into the pathophysiology of vascular diseases. Int J Mol Sci. 2019;20:E3675. doi: 10.3390/ijms20153675
86.
Fang X, An P, Wang H, Wang X, Shen X, Li X, Min J, Liu S, Wang F. Dietary intake of heme iron and risk of cardiovascular disease: a dose-response meta-analysis of prospective cohort studies. Nutr Metab Cardiovasc Dis. 2015;25:24–35. doi: 10.1016/j.numecd.2014.09.002
87.
Gao X, Liu X, Xu J, Xue C, Xue Y, Wang Y. Dietary trimethylamine N-oxide exacerbates impaired glucose tolerance in mice fed a high fat diet. J Biosci Bioeng. 2014;118:476–481. doi: 10.1016/j.jbiosc.2014.03.001
88.
Chen S, Henderson A, Petriello MC, Romano KA, Gearing M, Miao J, Schell M, Sandoval-Espinola WJ, Tao J, Sha B, et al. Trimethylamine N-Oxide Binds and activates PERK to promote metabolic dysfunction. Cell Metab. 2019;30:1141–1151.e5. doi: 10.1016/j.cmet.2019.08.021
89.
de Souza RJ, Mente A, Maroleanu A, Cozma AI, Ha V, Kishibe T, Uleryk E, Budylowski P, Schünemann H, Beyene J, et al. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies. BMJ. 2015;351:h3978. doi: 10.1136/bmj.h3978
90.
Witkowski M, Weeks TL, Hazen SL. Gut microbiota and cardiovascular disease. Circ Res. 2020;127:553–570. doi: 10.1161/CIRCRESAHA.120.316242
91.
Zhang X, Li Y, Yang P, Liu X, Lu L, Chen Y, Zhong X, Li Z, Liu H, Ou C, et al. Trimethylamine-N-oxide promotes vascular calcification through activation of NLRP3 (nucleotide-binding domain, leucine-rich-containing family, pyrin domain-containing-3) inflammasome and NF-κB (nuclear factor κB) signals. Arterioscler Thromb Vasc Biol. 2020;40:751–765. doi: 10.1161/ATVBAHA.119.313414
92.
Manson JE, Cook NR, Lee I-M, Christen W, Bassuk SS, Mora S, Gibson H, Albert CM, Gordon D, Copeland T, et al. Marine n−3 fatty acids and prevention of cardiovascular disease and cancer. New England Journal of Medicine. 2018;380:23–32
93.
Mozaffarian D, Wu JH. Omega-3 fatty acids and cardiovascular disease: effects on risk factors, molecular pathways, and clinical events. J Am Coll Cardiol. 2011;58:2047–2067. doi: 10.1016/j.jacc.2011.06.063