Effects of ketogenic and low‐carbohydrate diet on glucose and weight control in type 2 diabetes mellitus patients along with metabolome change

Abstract Objective Few trials have explored associations between diets, diet‐induced metabolomic changes and metabolic outcomes in type 2 diabetes mellitus (T2DM), especially weight‐stratified management. This study aimed to identify diet‐specific metabolomic signatures by defining ketogenic diet (KD) and low‐carbohydrate diet (LCD) applicability in T2DM patients with varying body mass index (BMI). Methods A total of 611 patients with T2DM were prescreened, and 505 eligible participants were included after exclusions. All participants received metformin and were allocated to either the LCD or KD group according to the study protocol and clinical dietary management process. BMI‐stratified medical phenotype analysis was performed, with clinical/serum profiles measured at baseline and 3 months. High‐throughput untargeted chromatography‐mass spectrometry analysed serum metabolome. Associations between metabolites and diet intake were examined and diet‐induced metabolomic changes versus diabetes prognosis were analysed via correlation. Results Of 611 prescreened participants, 505 eligible participants were included in the final analysis after completing the 12‐week intervention: Obese patients in the KD group had the most significant glycated haemoglobin (HbA1c) (8.18 ± 2.2 vs. 5.87 ± 0.98%) and weight (88.51 ± 15.8 vs. 80.5 ± 15.3 kg) reductions. Overweight patients benefited from both diets for weight and glucose control. Metabolomic analysis identified 2410 markers post‐intervention: KD group (279 vs. 259, up and downregulated); LCD group (300 vs. 288). Correlation regression: for the LCD group, weight negatively with histidinyl hydroxyproline; HbA1c positively with 4‐hydroxyphenylacetylglutamic acid. As for the KD group, HbA1c positively correlated with glutamyl isoleucine and negatively correlated with diethanolamine. Conclusion Both LCD and KD are effective for overweight and obese patients with T2DM. In the obese subgroup, LCD showed a numerically greater HbA1c reduction than KD, while both diets were suitable for overweight patients. For patients with normal BMI, LCD demonstrated no weight loss benefit. Metabolomic signatures sensitively reflect the effects of different diets and are closely associated with clinical metabolic improvements in T2DM, which is crucial for optimising dietary strategies. Trial Registration https://www.chictr.org.cn/ ChiCTR2500102119. Registration Date: 8 May, 2025