One-Size-Fits-All Models Don't Predict Postprandial Glucose Spikes

A personalized approach to diet-based glucose control may be most effective for reducing hyperglycemia, researchers suggested.

After normoglycemia individuals in the study ate the same food, the range of postprandial glycemic responses varied greatly, Helena Mendes-Soares, PhD, of the Mayo Clinic in Rochester, Minnesota, and colleagues reported in JAMA Network Open.

For example, after the 327 individuals in the study ate a standard meal consisting of a bagel and cream cheese, glycemic excursions ranged from just 6 mg/dL up to 94 mg/dL.

These participants, who wore continuous glucose monitors for 6 days, were fed four standard meals consumed as the first meal each day throughout the study, but were otherwise instructed to follow their normal diet. Each of the standardized meals was assessed for calorie and carbohydrate content. Along with this, stool samples were collected to analyze the individual microbiome composition.

Taking this a step further, Mendes-Soares and co-authors developed a novel model specifically designed to predict an individual's glucose response to food, which incorporated 72 factors. Some of these factors incorporated the individuals' microbiome features, meal content, and individual anthropometric factors.

The model also showed a high degree of accuracy, as the predicted and observed postprandial glycemic responses were highly correlated with each other, the researchers reported. Most interesting, however, they said, was that this model was even more accurate in predicting glycemic response compared with the "current standard-of-care approaches" based on calorie and carbohydrate content of food (R=0.62 for model vs R=0.34 for calories and R=0.40 for carbohydrates).

"The standard approach of counting carbohydrates and calories does not work as well because it considers only the characteristics of food. It fails to factor in the unique microbiome and lifestyle of each person," Mendes-Soares said in a statement.

Why aren't the current models as accurate? Study co-author Purna Kashyap, MBBS, also of the Mayo Clinic, explained that those models typically lump everything -- like carbohydrates and fats -- simply into one category. "With our individualized model, people no longer have to give up all foods within a certain category. It allows them to choose specific foods within certain categories that fit well with their microbiome."

Added senior study author Heidi Nelson, MD, also of the Mayo Clinic: "This study is the first critical step in defining and proving the value of a personalized diet. As a clinician, I have seen that my patients do not respond to the same foods the same way -- just like not all weight-loss diets work for all people the same," she said in a statement.

However, the researchers did point out several limitations to the study, mainly including barriers to the use of this model, one of which includes the need for patients to undergo fecal testing.

"Nevertheless, exploring the performance costs of simplified versions of our model containing only easily measurable clinical variables could help in making personalized nutrition accessible to larger populations," the team suggested.

Comment