The vegan vs. omnivore twin study: can controlling for gene-nutrient interactions uncover the best diet for heart health? Original paper

In this 8-week randomized controlled study, a vegan diet reduced body weight, LDL cholesterol levels, and insulin when compared to an omnivorous diet. These findings were strengthened because the study participants were identical twins.

The effects of a healthy vegan diet compared to a healthy omnivorous diet on cardiometabolic health in identical twins.

The primary outcome was low-density lipoprotein cholesterol (LDL-C). The secondary study outcomes included body weight, triglycerides, high-density lipoprotein cholesterol (HDL-C), blood glucose, insulin, trimethylamine N-oxide (TMAO), and serum vitamin B₁₂. Dietary intake data were collected as an exploratory outcome.

A total of 22 pairs of identical twins (44 participants in total; average age of 40; 34 women and 10 men).

In this 8-week randomized study, the participants were assigned to a healthy vegan diet or a healthy omnivorous diet. One twin was assigned to each dietary intervention.

The intervention was delivered in two 4-week phases: one in which meals were delivered to the participants and one in which meals were participant-provided.

Both diets were considered healthy as they emphasized the consumption of vegetables, fruits, and whole grains and discouraged the intake of added sugars and refined grains. In other words, the diets were similar except for the exclusion of animal-based foods in the vegan diet group (i.e., meat, poultry, fish, eggs, and dairy).

LDL-C decreased (improved) by 13.9 mg/dL in the vegan diet group compared to the omnivorous diet group.

In terms of secondary outcomes, the vegan diet group had lower insulin levels (−2.9 µIU/mL) and a lower body weight (−1.9 kg/4.2 lb) compared to the omnivorous diet group. Although HDL-C and vitamin B₁₂ were also lower in the vegan diet group, the differences were not statistically significant compared to the omnivore group.

In a sensitivity analysis in which three outlier TMAO values were removed, TMAO levels were lower in the vegan diet group compared to the omnivore diet group.

The effects of vegan and omnivorous diets on cardiometabolic risk factors in identical twins

Are plant-based diets “heart healthier” than omnivorous diets? Nutrition science has investigated this question for decades and, despite valiant research efforts, there is still no simple answer.

Observational and clinical studies suggest that vegan/vegetarian diets (hereafter referred to as “plant-based diets”) are associated with lower risks for ischemic heart disease^[1]; reduced body weight and reduced ApoB levels^[2]; and a lower risk of all-cause mortality, cancer, and cardiovascular disease.^[3] In fact, plant-based dietary patterns (i.e., lacto-ovo vegetarian, pescatarian, and vegan) were ranked among the healthiest by the American Heart Association (AHA) for aligning with the dietary guidelines for optimal cardiovascular health. The two highest-ranked diets were the Dietary Approaches to Stop Hypertension (DASH) and the Mediterranean diet.^[4] (Side note: the first author of the AHA Scientific Statement on popular dietary patterns was Dr. Christopher Gardner, the principal author of the current study).

But observational studies are limited in their ability to infer causation. In other words, people who eat plant-based diets might have better health outcomes than meat eaters, but is that solely due to what they’re eating? It could be, but confounding variables could also explain part of the association.

It’s possible that people who are more conscious about their health choose to adopt a vegan or vegetarian diet for personal or environmental reasons. Maybe these people are also more likely to engage in positive health behaviors like not smoking, moderating their alcohol intake, and exercising. Indeed, one population-based study concluded that vegetarians appeared more “health conscious” than nonvegetarians due to being more physically active and more likely to use supplements.^[5] On the other hand, most people eat meat in some form, and studies don’t usually account for meat type, what the meat is served with, and other behaviors that “meat eaters” engage in. Observational studies aren’t investigating a “healthy” omnivorous diet, for the most part.

For the above reason, randomized controlled studies are the best way to study nutrition interventions: assign two groups to two different diets for a set amount of time and observe what happens. At the end of the study, researchers have an answer as to which diet is “healthiest” or which diet is best for whatever outcome they’re studying … maybe. Even RCTs have their limitations.

Recent research has highlighted the importance of gene-nutrient interactions in determining the health effects of a diet. In other words, how a diet affects someone is a function not only of the nutrients comprising the diet but also how those nutrients interact with the person’s genes. This is called nutrigenomics.^[6]

To control for genetic influences, researchers would need a clone of a participant. Luckily, there’s nothing closer to a human clone than an identical twin. Identical twins share nearly the same DNA, have the same sex, and in many cases, experienced similar environments, making them the perfect participants for a unique scientific study in which genes and environment can be controlled for, allowing researchers to isolate the effects of their intervention.^[7]

The current study leveraged the Stanford Twin Registry to recruit the participant sample of 17 female and 5 male identical twin pairs — 79% of whom reported that they currently lived with their twin and 69% of whom reported “being similar” to their twin.

The use of identical twins increases confidence that any findings observed in the study weren’t due to a genetic quirk where some participants were “hardwired” to respond better to a vegan or omnivorous diet. It’s definitely a first-of-its-kind diet trial.

Some intricacies of each diet are worth pointing out. For one, the vegan diet group ate less protein than the omnivorous diet group — 58 grams per day compared to 91 grams per day — making up 14% and 20% of the diets’ calories, respectively. Both diets comprised about 41%–51% of calories from carbohydrates and about 36%–39% of calories from fat.

Vitamin B₁₂ intake was lower in the vegan diet group compared to the omnivore diet group, averaging 1.22 micrograms per day and 3.98 micrograms per day, respectively (this difference was not statistically significant). Not surprisingly, cholesterol intake was considerably lower in the vegan diet group (14.3 milligrams per day) than in the omnivore diet group (402 milligrams per day). Iron, however, was higher in the vegan diet group (18.11 mg/day) compared to the omnivore group (13.71 mg per day).

The biggest hole in this study, and one of the main points of criticism, was the lack of caloric control, which resulted in the vegan diet group consuming almost 200 kcal less per day than the omnivore diet group. The average caloric intake across both study phases was 1,658 kcal in the vegan diet group compared to 1,839 in the omnivore diet group. This calorie deficit explains the greater weight loss in the vegan diet group. Whether this weight loss came from fat mass or lean mass is unknown, as the researchers didn’t measure body composition. So while the study suggests a well-formulated vegan diet can help with weight loss, it doesn’t provide much insight into the effect of such a diet when weight loss isn’t part of the equation.

Additionally, the lack of statistically and clinically meaningful differences between the groups may have been the result of a few factors.

First, the participants were cardiometabolically healthy at baseline: they had a BMI of 26, a resting systolic and diastolic blood pressure of 124/75, a fasting blood glucose of 91 mg/dL, and a baseline LDL-C of 115. Because these metrics were not elevated, it means that there was likely less room for them to improve.

Second, the participants in both groups improved their diet quality — they increased their intake of vegetables and whole grains and reduced their intake of refined grains and sugar compared to their prestudy diets.

A final note concerns diet satisfaction. Though the objective health effects of any diet are important, so is the ability to sustain a diet in the long term. The vegan diet appeared at least marginally better for health, but how did the participants rate its acceptability?

Overall, satisfaction with the vegan diet was lower (3/5 points) than satisfaction with the omnivore diet (3.6/5 points). The lowest-scored components of the vegan diet were eating out (2.6/5 points), planning and preparation of food (2.7/5), and preoccupation with food (2.9/5). In the omnivore diet group, cost (3.1/5), planning and preparation of food (3.3/5), and preoccupation with food (3.3/5) rated lowest on the satisfaction scale. Furthermore, only 1 participant in the vegan diet group said they would “continue to closely follow all recommendations for my eating pattern” compared to 6 participants in the omnivore group who said the same thing.

Had the study been isocaloric (i.e., diets matched for calories), the outcome may have been different. However, that wasn’t the aim of the study — the authors designed this as a “free-living” trial to enhance its ecological validity. Furthermore, because this study was only 8 weeks in duration, it wasn’t possible to measure clinical outcomes like cardiovascular disease events or all-cause mortality. For those reasons and more, this study still leaves a few more questions than answers about which diet is the “healthiest”.

Dr. Christopher Gardner, the study’s principal investigator, receives funding from the Beyond Meat company. Although the study did not include Beyond Meat products specifically, plant-based meat alternatives are often a main component of vegan/vegetarian diets as a way for individuals to increase their intake of protein.

The secondary (e.g., insulin and body weight) outcomes weren’t adjusted for multiple comparisons, and so these results should be interpreted with caution.