Cardiometabolic Effects of Omnivorous vs Vegan Diets in Identical Twins

Key Points Question What are the cardiometabolic effects of a healthy plant-based (vegan) vs a healthy omnivorous diet among identical twins during an 8-week intervention? Findings In this randomized clinical trial of 22 healthy, adult, identical twin pairs, those consuming a healthy vegan diet showed significantly improved low-density lipoprotein cholesterol concentration, fasting insulin level, and weight loss compared with twins consuming a healthy omnivorous diet. Meaning The findings from this trial suggest that a healthy plant-based diet offers a significant protective cardiometabolic advantage compared with a healthy omnivorous diet.


Introduction
Plant-based diets have gained recent popularity not only for their lower environmental impact compared with an omnivorous dietary pattern but also for their health benefits. 1,2The most significant global health crises affecting our generation are noncommunicable diseases and climate change, which are both inextricably linked to diet, 3 and dietary patterns high in plants and low in animal foods can maximize health and environmental benefits. 4,5Plant-based diets contain a diverse family of dietary patterns, which encourage a reduced consumption of animal foods. 6Abundant evidence from observational and intervention studies [7][8][9][10][11][12][13] indicates that vegan diets are associated with improved cardiovascular health and decreased risk of cardiovascular disease, likely because of the higher daily consumption of vegetables and fruits, legumes, whole grains and nuts, and seeds compared with other different types of dietary patterns. 14vegan dietary pattern is typically lower in energy density but higher in fiber, vitamins, minerals, and phytonutrients compared with other dietary patterns. 15However, sometimes a vegan dietary pattern can limit specific nutrients, such as vitamin B 12 , iron, and calcium. 15,16Most studies 17,18 examining vegan diets have been epidemiologic examinations, with a few reported clinical studies. 19,20A confounding factor to consider in epidemiologic studies is the bias of self-decided vegans who may differ from nonvegans in factors that may influence diet and health. 21In addition, a poorly formulated vegan diet can include low-quality plant foods, such as refined carbohydrates and added sugars. 22To address these concerns, we designed a trial to compare the cardiometabolic effects of a healthy vegan diet with a healthy omnivorous diet, exposing both groups to vegetables, legumes, fruits, whole grains, nuts, and seeds.To control for genetic differences that might alter the cardiometabolic effects of diet, 23 we randomly assigned identical twins to follow the 2 diets for 8 weeks.
Stanford Twin Registry and randomized using computerized random-number generation by a statistician (K.M.C.) blinded to the intervention, delivery, or data collection.Adult twins 18 years or older willing to consume a plant-based (vegan) or omnivore diet for 8 weeks were included.We excluded participants who weighed 45.36 kg (100 lb) or less, had a body mass index (calculated as weight in kilograms divided by height in meters squared) of 40 or higher, had an LDL-C level of 190 mg/dL or higher (to convert to millimoles per liter, multiply by 0.0259), had a systolic blood pressure of 160 mm Hg or higher or diastolic blood pressure of 90 mm Hg or higher, or were pregnant.Individuals self-reported race and ethnicity for the purpose of demographic reporting.Inclusion and exclusion criteria have been previously published. 25

Dietary Intervention
The study consisted of two 4-week phases: delivered meals and self-provided meals.Participants were provided all no-cost meals for the first 4 study weeks by a nationwide meal delivery company (Trifecta Nutrition).It was expected that after 4 weeks of food delivery and health educator counseling that participants would understand the amounts and types of foods they should purchase and prepare to achieve maximum adherence to the diets when self-providing meals.
Research staff worked with Trifecta Nutrition to develop menu offerings to match a healthy vegan and omnivorous diet, which emphasized vegetables, fruits, and whole grains while limiting added sugars and refined grains.During the initial 4 weeks, meals were delivered once each week, with 7 days of breakfast, lunch, and dinner meals.Participants also purchased and consumed snacks to meet their energy requirements following guidance from health educators.
Guiding principles were reinforced: (1) choose minimally processed foods; (2) build a balanced plate with vegetables, starch, protein, and healthy fats; (3) choose variety within each food group; and (4) individualize these guidelines to meet preferences and needs (eAppendix in Supplement 2).
Although weight loss was not discouraged, our diet design did not include a prescribed energy restriction and was not intended to be a weight loss study.Participants were told to eat until they were satiated throughout the study.

Collection of Dietary Intake
Two types of dietary data were collected.For the primary reporting data, 3 unannounced 24-hour dietary recalls-a structured interview intended to capture detailed information about food and drink intakes-were administered within a 1-week window (2 weekdays and 1 weekend day) of each time point (baseline, week 4, and week 8).Data were collected via telephone by a registered dietitian (L.R.D.) using Nutrition Data System for Research (Nutrition Coordinating Center).For the secondary reporting data, participants were encouraged to log their food intake using the Cronometer app (Cronometer Pro, Nutrition Tracking Software for Professionals; Cronometer); these data were used by health educators for real-time guidance of participants.

Anthropometric and Metabolic Data
At 3 time points, participants visited the Stanford Clinical and Translational Research Unit after an overnight fast of 10 to 12 hours: baseline, 4 weeks (phase 1), and 8 weeks (phase 2).Blood draw and clinical measures were assessed using standard methods (eMethods in Supplement 2).Stool samples were collected for future analysis to examine changes to the gut microbiome (eg, microbial diversity), metabolites, inflammatory markers, and additional health factors.

Statistical Analysis
Descriptive statistics, mean (SD) or number (percentage), were used for continuous and categorical variables, respectively.Table 1 presents baseline summary statistics by study group.For the primary analysis, we investigated differences between groups in the change from baseline to week 8 for LDL-C between vegan and omnivorous diets among identical twins.Primary analysis included all available data.A linear mixed model was used and included fixed effects for diet and time (baseline as reference) and an interaction effect for diet × time and a random effect for twin pair to account for the correlation between identical twins (ie, random intercept allowed intercept to vary for each twin pair).A Wald test was used to evaluate a significant difference in diet at 8 weeks from baseline (interaction term).Finally, we present model estimates (95% CIs) for diet at 8 weeks.For each secondary outcome, we evaluated a statistical model similar to the primary model as described herein.
Analyses were completed using R Studio, version 2022.12.0 (Posit Software).A 2-sided P Յ .05 was considered to be statistically significant.No correction was applied for multiple comparisons, and secondary and exploratory analyses should be interpreted accordingly.

Primary Outcome
Participants receiving the vegan diet showed a mean (SD) decrease of 13.9 (5.8) mg/dL (95% CI, −25.3 to −2.4 mg/dL) in the unadjusted mean LDL-C level at 8 weeks from baseline compared with participants receiving the omnivorous diet (Table 2).As early as 4 weeks, we observed a significant decrease in mean LDL-C level among vegans compared with omnivores (eTable 20 in Supplement 2).
The percentage of change from baseline to 8 weeks in primary and secondary outcomes between vegan and omnivorous diet groups (Figure 2) showed a significant decrease in LDL-C level among the
Vegan participants had a significant mean (SD) decrease of −1.9 (0.7) kg in body weight (95% CI, −3.3 to −0.6 kg) from baseline to 8 weeks compared with participants on the omnivorous diet (P = .01)b Means (SEMs) are unadjusted.
c Primary and secondary outcomes fixed effects for diet and time (baseline as reference) and an interaction effect for diet (omnivore as reference) by time, a random effect for twin pair to account for the correlation between identical twins (ie, random intercept allowed intercept to vary for each twin pair), and a random effect for participant to account for correlation of longitudinal data.(Figure 2), although weight loss was observed for both diet groups.Vegans also experienced a larger but nonsignificant absolute median decrease in fasting high-density lipoprotein cholesterol, triglycerides, vitamin B 12 , glucose, and TMAO levels at 8 weeks from baseline compared with omnivores.

Sensitivity Analysis
Three outlier TMAO levels greater than 15 μM were noted: 2 at baseline and 1 at 8 weeks.After the outliers were eliminated, the TMAO level was significantly different between diet groups at 8 weeks: in this analysis, participants on the vegan diet showed a mean (SD) decrease of −2.1 (0.7) μM (95% CI, −3.5 to −0.7 μM) in the difference of TMAO from baseline to 8 weeks compared with participants on the omnivorous diet (eFigure 13 in Supplement 2).

Exploratory Analysis
Paired and unpaired 2-tailed t tests indicate minimal differences between statistical analysis approach (eTables 21 and 22 in Supplement 2).Participants receiving the omnivorous diet had nominally higher diet satisfaction at weeks 4 and 8 compared with vegan participants (eTable 13 in Supplement 2).Additional results are available in eResults and eTables 14 to 20 in Supplement 2.

Discussion
In this randomized clinical trial of healthy, adult identical twins, the 8-week change in LDL-C level-the primary outcome-was significantly lower for twins receiving the vegan diet compared with twins receiving the omnivorous diet.Insulin levels and weight were also significantly lower among the twins on the vegan diet from baseline to 8 weeks.Vegan-diet participants had total lower protein intake as a percentage of calories, lower dietary satisfaction, lower intake of dietary cholesterol, but higher intake of vegetable servings and intake of dietary iron.Vegans had lower intake of vitamin B 12 , yet serum vitamin B 12 levels were not statistically different than omnivores at 8 weeks, likely because of preserved stores. 27Long-term vegans are typically encouraged to take a cyanocobalamin (vitamin B 12 ) supplement.
Two factors may have limited our opportunity to observe additional differences between the study groups.First, participants in both diet groups were assigned to eat a healthy diet, usually healthier compared with their prestudy dietary pattern demonstrated by increased vegetable intake and decreased refined grains intake.Even the omnivorous participants improved their diet quality during the 8-week intervention (eg, increased vegetables and whole grain intake and decreased added sugars and refined grains).Second, within both groups, potential differences in clinical end point changes may have been blunted because participants were healthy at baseline.For example, participants' mean baseline LDL-C level was 114 mg/dL, 26 leaving minimal room for participants to improve through diet alone.Nonetheless, we observed significant improvements in 3 clinical outcomes (LDL-C, insulin, and weight) among the vegan participants.
Our results corroborate a previous finding showing that eating a vegan diet can improve cardiovascular health. 28A larger body of evidence from randomized clinical trials suggests that vegetarian and other plant-based dietary patterns lower weight [29][30][31] and improve lipid management, 30,32,33 glucose metabolism, 33,34 blood pressure, [35][36][37] and cardiometabolic health. 38r results also mirror a recently completed 2-year dietary intervention trial among African Americans randomized to a vegan or low-fat omnivorous diet, finding improvements in body weight and cardiovascular disease risk factors. 39vel to this study was our population of identical twins, a valuable resource in scientific research that provided a unique opportunity to investigate the effects of a dietary intervention while controlling for genetic and environmental factors, 40 influences that can significantly impact health outcomes, including body weight, cardiovascular health, and metabolic function. 40,41Because identical twins have nearly identical DNA and many shared experiences (eg, upbringing, geographic region growing up, and similar exposure to other variables), observed differences in health outcomes after adoption of different dietary patterns can largely be attributed to the diet itself.
We were surprised that TMAO concentrations did not significantly differ between diets at 8 weeks because of the higher meat content in the omnivorous diet and of the meat TMAO precursors choline and carnitine. 42,43Although some studies 44,45 report a positive association between the concentration of serum TMAO and development of cardiovascular disease, whether TMAO is a bystander or mediator of disease remains unknown.In a sensitivity analysis that removed 3 TMAO outlier participants, lower TMAO levels were found in the vegan participants.Prior research 42,43 has suggested that vegans have lower TMAO levels than meat or fish eaters because of the TMAO precursors choline and carnitine in animal products.In a recent crossover dietary trial (Study With Appetizing Plantfood-Meat Eating Alternative Trial [SWAP-MEAT]), 46 participants consuming plantbased alternative meat vs animal meat had significantly lower TMAO concentrations.In addition to our 3 TMAO outliers, we observed variability among participants in TMAO concentration changes.
Further investigation is needed on TMAO as a risk factor for cardiovascular disease and the association of dietary choline and carnitine vs fish with serum TMAO concentrations.
A recent meta-epidemiologic study 47 examining dietary recommendations from current clinical practice guidelines recommends diets rich in unrefined plant foods and low in refined and animalbased foods.Clinical practice guidelines from the American Heart Association recommend that practitioners encourage patients to choose healthy sources of protein, mostly from plants, to promote cardiovascular health. 11,48Additionally, Dietary Guidelines for Americans, 2020-2025 49 includes a healthy, vegetarian-style dietary pattern that can be adopted for improved health and chronic disease prevention.Although our findings suggest that vegan diets offer a protective cardiometabolic advantage compared with a healthy, omnivorous diet, excluding all meats and/or dairy products may not be necessary because research 22,50 suggests that cardiovascular benefits can be achieved with modest reductions in animal foods and increases in healthy plant-based foods compared with typical diets.We believe lower dietary satisfaction in the vegan group may have been attributable to the strictness of the vegan diet, creating more barriers for people to follow the vegan diet guidelines.Some people may find a less restrictive diet preferable for LDL-C-lowering effects.
Future studies assessing health benefits of less strict plant-based diets will be necessary to assess these benefits, especially in a study model limiting additional biases (eg, in twins).Within a clinical setting, patients should be supported in choosing a dietary pattern that fits their needs and preferences. 41,51Clinicians should allow patients to make informed choices that support them to choose which dietary approach is most suitable for them.At a population level, wider adoption of a culturally appropriate dietary pattern that is higher in plant foods and lower in animal foods can promote health and environmental benefits. 3,4,10,52

Strengths and Limitations
Several aspects of our design and implementation were strengths.First, enrolling identical twins was beneficial because we were able to eliminate the confounding influences of age, sex, and genetic factors that may affect clinical outcomes.Identical twins often share a similar environment and lifestyle, reducing environmental factors on the study results.Second, the initial 4-week period of food delivery facilitated participants' high adherence to the diet, whereas the latter 4 weeks of selfprovided foods increased generalizability.Third, we used LDL-C, a well-established cardiometabolic clinical value, as the primary outcome. 26Fourth, we assessed an extensive set of well-studied secondary clinical outcomes to evaluate overall cardiometabolic health.Fifth, diet data collection using the state-of-the-art Nutrition Data System for Research allowed us to assess and report on adherence-an important metric in free-living trials 53 -and compare macronutrient and micronutrient intakes.Sixth, previous trials 11,13,31,50,54,55

Figure 2 .
Figure 2. Median Change From Baseline to 8 Weeks in Primary and Secondary Outcomes Between Vegan and Omnivorous Diet Arms

48 For
primary and secondary outcomes, percent change and P values are presented.A Wald test was used to evaluate a significant difference in diet at 8 weeks from baseline (interaction term).Error bars indicate IQRs.HDL-C indicates high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; and TMAO, trimethylamine N-oxide.

Table 2 .
Cardiovascular Health Outcomes at the End of 8 Weeks and Main Effect Model Estimates for Primary and Secondary Outcome Analyses Abbreviations: HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TMAO, trimethylamine N-oxide.SI conversion factors: To convert HDL-C and LDL-C to mmol/L, multiply by 0.0259; triglycerides to mmol/L, multiply by 0.0113; glucose to mmol/L, multiply by 0.0555; insulin to pmol/L, multiply by 6.945; vitamin B 12 to pmol/L, multiply by 0.7378.a All laboratory data are fasting values from plasma (lipids, glucose, insulin, and vitamin B 12 ) or serum (TMAO) specimens.
Fat Distribution by Phase and Diet eTable 3. Fat Distribution by Phase and Diet eFigure 4. Grain Distribution by Phase and Diet eTable 4. Grain Distribution by Phase and Diet eFigure 5. Protein Distribution by Phase and Diet eTable 5. Protein Distribution by Phase and Diet eFigure 6. Carbohydrate Distribution by Phase and Diet eTable 6. Carbohydrate Distribution by Phase and Diet eFigure 7. Dietary Cholesterol Distribution by Phase and Diet eTable 7. Dietary Cholesterol Distribution by Phase and Diet eFigure 8. Dietary Vitamin B12 Distribution by Phase and Diet eTable 8. Dietary Vitamin B12 Distribution by Phase and Diet eFigure 9. Dietary Iron Distribution by Phase and Diet eTable 9. Dietary Iron Distribution by Phase and Diet eFigure 10.Vegetable Servings Distribution by Phase and Diet eTable 10.Vegetable Servings Distribution by Phase and Diet eFigure 11.Animal-Based Protein Distribution by Phase and Diet eTable 11.Animal-Based Protein Distribution by Phase and Diet eFigure 12. Meat Alternatives Distribution by Phase and Diet eTable 12. Meat Alternatives Distribution by Phase and Diet eTable 13.Diet Satisfaction (D-Sat28) of Participants by Diet Assignment and Phase eTable 14.Self-Efficacy to Plan, Shop, Cook, and Choose Meals by Diet Assignment and Phase eTable 15.Diet Preferences of Participants by Diet Assignment eTable 16.Perceptions of Delivered, Pre-Prepared Meals by Diet Assignment eTable 17.Barriers to Adherence to Study Eating Patterns by Diet Assignment eTable 18. Factors that Improve Dietary Adherence and Sustainability by Diet Assignment eTable 19.Average Self-Rated Dietary Adherence by Diet Assignment and Phase eTable 20.Cardiovascular Health Outcomes at the End of 4 Weeks and Main Effect Model Estimates, Standard Errors, and 95% Confidence Interval for Primary and Secondary Outcome Analysis eFigure 13.Sensitivity Analysis of the Change in Trimethylamine N-Oxide (TMAO) With Three Outliers Removed, (Mean ± SE) eTable 21.Paired T-Tests of Cardiovascular Health Outcomes at the End of 8 Weeks eTable 22. Unpaired T-Tests of Cardiovascular Health Outcomes at the End of 8 Weeks eTable 23.Average Macronutrient and Micronutrient Composition of Trifecta Food Delivery Meals by Meal Type and Diet Assignment eReferences 11ve reported similar metabolic and weight loss benefits of vegan diets yet tended to focus on very low-fat vegan diets, study populations with diabetes or overweight, and comparison diets with limited attention to equipoise.Novelties of the current trial were the use of a more moderate-and higher-fat vegan diet (unsaturated fat),11the