Personalized Diets by Prediction of Glycemic Responses Improve Glycemic Control in Subjects with Newly Diagnosed T2D

Dietary modifications are crucial for the management of newly-diagnosed type-2 diabetes (T2D) and prevention of its health complications, but many patients fail to achieve clinical goals with diet alone. We previously developed a machine-learning algorithm for predicting personalized postprandial glucose responses (PPGR) to meals using clinical and gut microbiome features, and showed that dietary interventions based on this algorithm successfully lowered PPGRs in adults with prediabetes. Here, we sought to evaluate the clinical efficacy of our algorithm-based diet in subjects with newly-diagnosed T2D. We performed a short-term randomized controlled crossover trial and compared the effects of an algorithm-based personalized postprandial-targeting (PPT) diet, to those of a commonly used Mediterranean (MED) diet on glucose levels in 23 subjects with newly-diagnosed T2D. Average PPGR, glucose fluctuations, daily time of glucose levels >140 mg/dl and fructosamine levels decreased significantly more during the PPT intervention compared to the MED intervention. We further evaluated the long-term clinical effects of the PPT diet in 16 of the participants by an additional 6-month PPT intervention, and found significant improvements in multiple clinical parameters, including HbA1c (mean±SD, -0.39±0.48%, p<0.001), fasting glucose (-16.4±24.2 mg/dl, p=0.02), fasting insulin (-2.3±4.0 MCU/ml, p=0.04), triglycerides (-49±46 mg/dl, p<0.001), body weight (-3±3.5 kg, p=0.005), body fat% (-2.5±3%, p=0.005) and waist circumference (-4.7±3.7 cm, p=0.001). Importantly, 61% of the participants exhibited diabetes remission at the end of the intervention, as measured by HbA1c. Finally, some of the improvements in clinical outcomes were accompanied by significant alterations to the gut microbiome composition. These findings may be valuable for the design of future studies in larger cohorts and may have implications for dietary advice in clinical practice. Disclosure M. S. Rein: None. O. Ben-yacov: None. A. Godneva: None. S. Shilo: None. S. Zelber-sagi: None. E. Elinav: Consultant; Self; BiomX, DayTwo. E. Segal: Consultant; Self; DayTwo.