User login
LOS ANGELES – .
“The reason we got interested in nutrition in general is for its important role in health and disease, but also because, reading the literature on nutrition in general, it seemed that the science was relatively poor,” Eran Segal, PhD, said at the World Congress on Insulin Resistance, Diabetes & Cardiovascular Disease. “As a testament, you can see how frequently dietary recommendations for the public are changed. For example, 30 years ago, the cover of Time magazine said that eating cholesterol in the diet is very bad for you. Fifteen years later, Time magazine said that some cholesterol is actually good for you. There are other questions, like should you be eating dairy products? I think it shows that we have a poor understanding of what healthy nutrition is for human individuals. That’s why we wanted to start a study which would collect the right amount and the right kind of data to try to answer the question of what is a healthy diet for human individuals.”
In what is believed to be the first study of its kind, Dr. Segal, professor of computer science and applied mathematics at the Weizmann Institute of Science in Rehovot, Israel, and his associates recruited 1,000 individuals and asked them to wear a continuous glucose monitor (CGM) for 1 week (Cell 2015;163[5]:1079-94). For the study, known as The Personalized Nutrition Project, participants were asked to log everything they ate into a mobile app the researchers developed. “They would select a meal from a database of 10,000 foods,” Dr. Segal explained. “Each meal has full nutritional value so at the end of the study, we had about 50,000 meals that we had measurements of postprandial glucose response to, coupled with full nutritional values.” They also collected a comprehensive profile of individuals, which included body measurements, blood tests, medical background, food frequency questionnaires, and a measurement of the microbiome by both 16S rRNA sequencing and shotgun metagenomics.
For the first part of the study, researchers supplied a breakfast to all participants: either bread, bread with butter, glucose, or fructose, in each case 50 g of available carbohydrates. “The participants were asked to consume these the morning after the night fast,” Dr. Segal said. “This allowed us to compare how the same individual responds to eating the exact same food versus how different individuals respond to eating the same food.” The researchers found that, when the same person ate the same meal on 2 different days, the glucose response was highly reproducible. In contrast, different people had widely different postmeal glucose responses to identical meals. “Some individuals responded most highly to glucose; others responded most highly to bread,” Dr. Segal said. “There was about 10% of individuals who responded to bread and butter, compared to the other test foods. These results mean that any universal diet is going to have limited efficacy in its ability to balance blood glucose levels, because some foods will spike glucose levels in one person but not in another person. It also means that the concepts we’ve been using like the glycemic index are also going to have limited efficacy.”
Next, the researchers aimed to determine what factors influence the variability in people’s responses to the same food. “We found many different correlations between the various blood markers and physical measurements that we obtained, but what was most novel was the variability in postmeal glucose response across people associated with microbiota composition and function,” Dr. Segal said. From this, he and his colleagues developed a machine-learning algorithm that integrates blood parameters, dietary habits, anthropometrics, physical activity, and gut microbiota. Using this algorithm, the prediction accuracy of personalized glucose responses achieved an r value of 0.68, which explains about 50% of the variability. For the final component of the study, the researchers randomized 26 participants to one of five dietary arms and followed for 1 week by continuous glucose monitoring. They were able to demonstrate that personally tailored diets lower the postprandial glucose response.
As a follow-up to this work, Dr. Segal and his associates enrolled 200 people with an hemoglobin A1c between 5.7% and 6.5% into the Personalized Nutrition Project for Prediabetes (PNP3) study, which investigates whether personalized diet intervention will improve postprandial blood glucose levels and other metabolic health factors in individuals with prediabetes, compared with the standard Mediterranean-style low-fat diet (NCT03222791). Participants were randomized to 6 months of standard of care following Dietary Guidelines for Americans 2015-2020, Eighth Edition, or to an algorithm diet. Primary outcomes are reduction in average glucose levels and evaluation of the total daily time of plasma glucose levels were below 140 mg/dL. Participants wore the continuous glucose monitor for the entire 6 months of intervention. “I don’t think this was ever done before,” he said. “We’re also looking at secondary metabolic endpoints and exploratory endpoints such as changes in the microbiome. We’re asking people to log everything they eat for the entire 6 months of intervention. It gives us a lot of power in terms assessing compliance. It’s an immense amount of data.”
Evaluation of the data are not yet complete, but interim results are promising. For example, he discussed results from one study participant on the algorithm diet. “Across 1 month, this person was able to entirely reduce the peaks in glucose levels and dramatic reductions in the time above 140 mg/dL in the 6-month treatment period,” said Dr. Segal, who is one of the study’s principal investigators. He disclosed that he is a paid consultant to DayTwo.
LOS ANGELES – .
“The reason we got interested in nutrition in general is for its important role in health and disease, but also because, reading the literature on nutrition in general, it seemed that the science was relatively poor,” Eran Segal, PhD, said at the World Congress on Insulin Resistance, Diabetes & Cardiovascular Disease. “As a testament, you can see how frequently dietary recommendations for the public are changed. For example, 30 years ago, the cover of Time magazine said that eating cholesterol in the diet is very bad for you. Fifteen years later, Time magazine said that some cholesterol is actually good for you. There are other questions, like should you be eating dairy products? I think it shows that we have a poor understanding of what healthy nutrition is for human individuals. That’s why we wanted to start a study which would collect the right amount and the right kind of data to try to answer the question of what is a healthy diet for human individuals.”
In what is believed to be the first study of its kind, Dr. Segal, professor of computer science and applied mathematics at the Weizmann Institute of Science in Rehovot, Israel, and his associates recruited 1,000 individuals and asked them to wear a continuous glucose monitor (CGM) for 1 week (Cell 2015;163[5]:1079-94). For the study, known as The Personalized Nutrition Project, participants were asked to log everything they ate into a mobile app the researchers developed. “They would select a meal from a database of 10,000 foods,” Dr. Segal explained. “Each meal has full nutritional value so at the end of the study, we had about 50,000 meals that we had measurements of postprandial glucose response to, coupled with full nutritional values.” They also collected a comprehensive profile of individuals, which included body measurements, blood tests, medical background, food frequency questionnaires, and a measurement of the microbiome by both 16S rRNA sequencing and shotgun metagenomics.
For the first part of the study, researchers supplied a breakfast to all participants: either bread, bread with butter, glucose, or fructose, in each case 50 g of available carbohydrates. “The participants were asked to consume these the morning after the night fast,” Dr. Segal said. “This allowed us to compare how the same individual responds to eating the exact same food versus how different individuals respond to eating the same food.” The researchers found that, when the same person ate the same meal on 2 different days, the glucose response was highly reproducible. In contrast, different people had widely different postmeal glucose responses to identical meals. “Some individuals responded most highly to glucose; others responded most highly to bread,” Dr. Segal said. “There was about 10% of individuals who responded to bread and butter, compared to the other test foods. These results mean that any universal diet is going to have limited efficacy in its ability to balance blood glucose levels, because some foods will spike glucose levels in one person but not in another person. It also means that the concepts we’ve been using like the glycemic index are also going to have limited efficacy.”
Next, the researchers aimed to determine what factors influence the variability in people’s responses to the same food. “We found many different correlations between the various blood markers and physical measurements that we obtained, but what was most novel was the variability in postmeal glucose response across people associated with microbiota composition and function,” Dr. Segal said. From this, he and his colleagues developed a machine-learning algorithm that integrates blood parameters, dietary habits, anthropometrics, physical activity, and gut microbiota. Using this algorithm, the prediction accuracy of personalized glucose responses achieved an r value of 0.68, which explains about 50% of the variability. For the final component of the study, the researchers randomized 26 participants to one of five dietary arms and followed for 1 week by continuous glucose monitoring. They were able to demonstrate that personally tailored diets lower the postprandial glucose response.
As a follow-up to this work, Dr. Segal and his associates enrolled 200 people with an hemoglobin A1c between 5.7% and 6.5% into the Personalized Nutrition Project for Prediabetes (PNP3) study, which investigates whether personalized diet intervention will improve postprandial blood glucose levels and other metabolic health factors in individuals with prediabetes, compared with the standard Mediterranean-style low-fat diet (NCT03222791). Participants were randomized to 6 months of standard of care following Dietary Guidelines for Americans 2015-2020, Eighth Edition, or to an algorithm diet. Primary outcomes are reduction in average glucose levels and evaluation of the total daily time of plasma glucose levels were below 140 mg/dL. Participants wore the continuous glucose monitor for the entire 6 months of intervention. “I don’t think this was ever done before,” he said. “We’re also looking at secondary metabolic endpoints and exploratory endpoints such as changes in the microbiome. We’re asking people to log everything they eat for the entire 6 months of intervention. It gives us a lot of power in terms assessing compliance. It’s an immense amount of data.”
Evaluation of the data are not yet complete, but interim results are promising. For example, he discussed results from one study participant on the algorithm diet. “Across 1 month, this person was able to entirely reduce the peaks in glucose levels and dramatic reductions in the time above 140 mg/dL in the 6-month treatment period,” said Dr. Segal, who is one of the study’s principal investigators. He disclosed that he is a paid consultant to DayTwo.
LOS ANGELES – .
“The reason we got interested in nutrition in general is for its important role in health and disease, but also because, reading the literature on nutrition in general, it seemed that the science was relatively poor,” Eran Segal, PhD, said at the World Congress on Insulin Resistance, Diabetes & Cardiovascular Disease. “As a testament, you can see how frequently dietary recommendations for the public are changed. For example, 30 years ago, the cover of Time magazine said that eating cholesterol in the diet is very bad for you. Fifteen years later, Time magazine said that some cholesterol is actually good for you. There are other questions, like should you be eating dairy products? I think it shows that we have a poor understanding of what healthy nutrition is for human individuals. That’s why we wanted to start a study which would collect the right amount and the right kind of data to try to answer the question of what is a healthy diet for human individuals.”
In what is believed to be the first study of its kind, Dr. Segal, professor of computer science and applied mathematics at the Weizmann Institute of Science in Rehovot, Israel, and his associates recruited 1,000 individuals and asked them to wear a continuous glucose monitor (CGM) for 1 week (Cell 2015;163[5]:1079-94). For the study, known as The Personalized Nutrition Project, participants were asked to log everything they ate into a mobile app the researchers developed. “They would select a meal from a database of 10,000 foods,” Dr. Segal explained. “Each meal has full nutritional value so at the end of the study, we had about 50,000 meals that we had measurements of postprandial glucose response to, coupled with full nutritional values.” They also collected a comprehensive profile of individuals, which included body measurements, blood tests, medical background, food frequency questionnaires, and a measurement of the microbiome by both 16S rRNA sequencing and shotgun metagenomics.
For the first part of the study, researchers supplied a breakfast to all participants: either bread, bread with butter, glucose, or fructose, in each case 50 g of available carbohydrates. “The participants were asked to consume these the morning after the night fast,” Dr. Segal said. “This allowed us to compare how the same individual responds to eating the exact same food versus how different individuals respond to eating the same food.” The researchers found that, when the same person ate the same meal on 2 different days, the glucose response was highly reproducible. In contrast, different people had widely different postmeal glucose responses to identical meals. “Some individuals responded most highly to glucose; others responded most highly to bread,” Dr. Segal said. “There was about 10% of individuals who responded to bread and butter, compared to the other test foods. These results mean that any universal diet is going to have limited efficacy in its ability to balance blood glucose levels, because some foods will spike glucose levels in one person but not in another person. It also means that the concepts we’ve been using like the glycemic index are also going to have limited efficacy.”
Next, the researchers aimed to determine what factors influence the variability in people’s responses to the same food. “We found many different correlations between the various blood markers and physical measurements that we obtained, but what was most novel was the variability in postmeal glucose response across people associated with microbiota composition and function,” Dr. Segal said. From this, he and his colleagues developed a machine-learning algorithm that integrates blood parameters, dietary habits, anthropometrics, physical activity, and gut microbiota. Using this algorithm, the prediction accuracy of personalized glucose responses achieved an r value of 0.68, which explains about 50% of the variability. For the final component of the study, the researchers randomized 26 participants to one of five dietary arms and followed for 1 week by continuous glucose monitoring. They were able to demonstrate that personally tailored diets lower the postprandial glucose response.
As a follow-up to this work, Dr. Segal and his associates enrolled 200 people with an hemoglobin A1c between 5.7% and 6.5% into the Personalized Nutrition Project for Prediabetes (PNP3) study, which investigates whether personalized diet intervention will improve postprandial blood glucose levels and other metabolic health factors in individuals with prediabetes, compared with the standard Mediterranean-style low-fat diet (NCT03222791). Participants were randomized to 6 months of standard of care following Dietary Guidelines for Americans 2015-2020, Eighth Edition, or to an algorithm diet. Primary outcomes are reduction in average glucose levels and evaluation of the total daily time of plasma glucose levels were below 140 mg/dL. Participants wore the continuous glucose monitor for the entire 6 months of intervention. “I don’t think this was ever done before,” he said. “We’re also looking at secondary metabolic endpoints and exploratory endpoints such as changes in the microbiome. We’re asking people to log everything they eat for the entire 6 months of intervention. It gives us a lot of power in terms assessing compliance. It’s an immense amount of data.”
Evaluation of the data are not yet complete, but interim results are promising. For example, he discussed results from one study participant on the algorithm diet. “Across 1 month, this person was able to entirely reduce the peaks in glucose levels and dramatic reductions in the time above 140 mg/dL in the 6-month treatment period,” said Dr. Segal, who is one of the study’s principal investigators. He disclosed that he is a paid consultant to DayTwo.
EXPERT ANALYSIS FROM WCIRDC 2018