No significant difference in diabetes management

In a recent study published in The American Journal of Clinical Nutrition, researchers compare changes in hemoglobin A1c (HbA1c) and glycemic variability in prediabetic individuals or those with moderately managed type 2 diabetes to understand the impacts of two calorie-restricted weight loss diets.

Study: A randomized clinical trial comparing low-fat versus precision nutrition-based diets for weight loss: impact on glycemic variability and HbA1c. Image Credit: Roman Chazov / Shutterstock.com

Background

Increased postprandial glucose response (PPGR) and the resultant oxidative damage are associated with a higher risk of type 2 diabetes, cardiovascular disease, cancer, liver disease, obesity, and associated mortality. Conventional strategies for PPGR management are through diets with low glycemic loads, such as those with low carbohydrates, and more soluble dietary fibers, such as whole plant foods.

However, standard diets with low glycemic load or low carbohydrate content have had mixed or negative results in clinical trials, as they do not consider individual variability in PPGR to the same foods.

About the study

In the present study, researchers used a machine learning-based personalized algorithm devised as part of the Personal Nutrition Project (PNP). PNP was trained on a data set comprising metagenome profiles of gut-microbiome-related glycemic responses and tolerance to predict individual PPGR.

The derived model also considered other data such as interstitial glucose measurements, the nutritional information of time-stamped meals, and characteristics of participants such as sleep, physical activity, hunger, stress levels, and HbA1c values.

A personalized diet based on the PNP algorithm was compared to a standardized low-fat diet in this randomized clinical trial called the Personal Diet Study by examining changes in HbA1c and glycemic variability during the intervention period. The study recruited adults between the ages of 18 and 80 years who were prediabetic or had type 2 diabetes managed either through lifestyle changes alone or with metformin along with lifestyle, had comorbid obesity or overweight, and an estimated glomerular filtration rate greater than 60 ml/min/1.73 m².

Participants were randomly assigned to the standardized or personalized diet. Stool and blood samples were obtained to predict the PPGRs using the PNP algorithm.

Both groups were provided behavioral counseling for weight loss and monitored their diets using an app on their smartphones. Individuals in the personalized diet group were provided feedback through the smartphone app to reduce PPGR. Data on continuous glucose monitoring were collected at baseline and three and six months.

The stool samples were used for microbiome analysis, which was then correlated with data on anthropometrics, health variables, sociodemographic information, and HbA1c to determine the personalized meal scores. The primary measured outcomes included mean amplitude of glycemic excursions (MAGE) and HbA1c values, while secondary outcomes included mean continuous glucose monitor readings, continuous overall net glycemic action (CONGA), coefficient of variation, and standard deviation.

Study findings

The personalized diet designed by the PNP algorithm did not result in any significant changes in HbA1c or glycemic variability as compared to the standardized diet.

The MAGE value decreased for the standardized diet group by 0.83 mg/dL each month. Comparatively, for the personalized diet group, the reduction in MAGE was at the rate of 0.79 mg/dL each month. The changes in HbA1c also showed similar trends for the two groups.

In addition to the coefficient of variation for the continuous glucose monitor glucose measures, other measures of glycemic variation and HbA1c decreased for both groups. Between the two groups, no differences were observed, even when the analysis was stratified by sex.

Changes in HbA1c every month were 0.02% and 0.01% for the standardized and personalized diet groups, respectively. The differences in the values between the groups were not statistically significant.

Some of the limitations of the study included the inability to generalize the findings across racial and ethnic groups, as the study population was primarily female and White. Furthermore, since the PNP algorithm was designed for the Israeli population and validated on a predominantly White sample of the United States population; thus, the results of the algorithm for other racial and ethnic groups could differ. Difficulties with using the app could also have resulted in a decrease in adherence to the self-monitored diet.

Conclusions

A personalized PNP algorithm-designed weight loss diet did not result in a significant difference in HbA1c or glycemic variability values as compared to the standardized weight-loss diet among individuals in prediabetic stages or with managed type 2 diabetes.