The Personalized Nutrition Project found that study participants had strikingly different responses to identical foods. In participant 445 (top), blood sugar levels rose sharply after eating bananas but not after cookies of the same amount of calories. The opposite occurred in participant 644 (bottom). [Weizmann Institute of Science]
The Personalized Nutrition Project found that study participants had strikingly different responses to identical foods. In participant 445 (top), blood sugar levels rose sharply after eating bananas but not after cookies of the same amount of calories. The opposite occurred in participant 644 (bottom). [Weizmann Institute of Science]

This Thanksgiving, before you load your plate with turkey, sweet potatoes, stuffing, and all the usual holiday fare, you might want to reflect on the findings of the Personalized Nutrition Project. This study found that people who consumed identical meals showed huge differences in the rise of their blood sugar levels. This finding suggests that people would be more likely to stay healthy if they were to shun universal dietary advice, and instead embrace personalized diets.

The Personalized Nutrition Project focused on blood sugar because elevated levels are a major risk factor for diabetes, obesity, and metabolic syndrome. It also recognized that existing dietary methods often fail to control blood sugar adequately.

To account for the risk of elevated blood sugar, doctors and nutritionists rely on a decades-old standard, the glycemic index (GI), which ranks foods based on how they affect blood sugar level. This approach, however, is based on studies that average how small groups of people respond to various foods.

Hoping to find a more nuanced approach, scientists at the Weizmann Institute decided to undertake what they characterized as the largest investigation of its kind. The Weizmann scientists, led by computer scientist Eran Segal, Ph.D., and immunologist Eran Elinav, Ph.D., continuously monitored blood sugar levels in 800 people for a week. Ultimately, the scientists assessed the responses of study participants to more than 46,000 meals.

The results of this work appeared November 19 in the journal Cell, in an article entitled, “Personalized Nutrition by Prediction of Glycemic Responses.” It found high variability in the response to identical meals, supporting the idea that universal dietary recommendations have limited utility.

“We devised a machine-learning algorithm that integrates blood parameters, dietary habits, anthropometrics, physical activity, and gut microbiota measured in this cohort and showed that it accurately predicts personalized postprandial glycemic response to real-life meals,” wrote the authors of the Cell article. “[A] blinded randomized controlled dietary intervention based on this algorithm resulted in significantly lower postprandial responses and consistent alterations to gut microbiota configuration.”

The study was unique not only in its scale, but also in its inclusion of the analysis of gut microbes, collectively known as the microbiome. In addition, the study asked participants to record everything they ate, as well as such lifestyle factors as sleep and physical activity.

Taking these multiple factors into account, the scientists generated an algorithm for predicting individualized response to food based on the person's lifestyle, medical background, and the composition and function of his or her microbiome. In a follow-up study of another 100 volunteers, the algorithm successfully predicted the rise in blood sugar in response to different foods, demonstrating that it could be applied to new participants. The scientists were able to show that lifestyle also mattered. The same food affected blood sugar levels differently in the same person, depending, for example, on whether its consumption had been preceded by exercise or sleep.

In the final stage of the study, the scientists designed a dietary intervention based on their algorithm; this was a test of their ability to prescribe personal dietary recommendations for lowering blood glucose level responses to food. Volunteers were assigned a personalized “good” diet for one week, and a “bad” diet—also personalized—for another. Both good and bad diets were designed to have the same number of calories, but they differed between participants. Thus, certain foods in one person's “good” diet were part of another's “bad” diet.

The “good” diets indeed helped to keep blood sugar at steadily healthy levels, whereas the “bad” diets often induced spikes in glucose levels—all within just one week of intervention. Moreover, as a result of the “good” diets, the volunteers experienced consistent changes in the composition of their gut microbes, suggesting that the microbiome may be influenced by the personalized diets while also playing a role in participants' blood sugar responses.

“After seeing this data, I think about the possibility that maybe we're really conceptually wrong in our thinking about the obesity and diabetes epidemic,” said Dr. Segal. “The intuition of people is that we know how to treat these conditions, and it's just that people are not listening and are eating out of control—but maybe people are actually compliant but in many cases we were giving them wrong advice.”

“Measuring such a large cohort without any prejudice really enlightened us on how inaccurate we all were about one of the most basic concepts of our existence, which is what we eat and how we integrate nutrition into our daily life,” explained Dr. Elinav. “In contrast to our current practices, tailoring diets to the individual may allow us to utilize nutrition as means of controlling elevated blood sugar levels and its associated medical conditions.”

Previous articleNearly Indestructible Tardigrade Holds Record for Most Foreign DNA
Next articleMeaningful Endpoints for Immuno-Oncology Trials