Pre-diabetes, characterized by elevated blood glucose levels, significantly increases the risk of type 2 diabetes and related comorbidities. Approximately 7.5% of the global population is affected, posing a major health concern. Diet plays a crucial role in the development of hyperglycemia and pre-diabetes; poor nutrition contributes to inflammation and pancreatic beta-cell damage. However, individual responses to diet vary significantly, highlighting the need for personalized approaches. Previous research utilized a machine-learning algorithm to predict personalized postprandial glycemic responses to real-life meals, leading to the development of a personalized postprandial glucose-targeting diet (PPT). The gut microbiome, known to mediate the relationship between diet, metabolism, and immunity, and the oral microbiome, which is linked to hyperglycemia through local and systemic inflammation, are both potential targets for dietary interventions. This study aimed to assess the impact of PPT and MED diets on the oral and gut microbiome, metabolites, and cytokines in pre-diabetic individuals to better understand their potential in pre-diabetes management and clarify the microbiome's mediating role.

Existing literature highlights the critical role of diet in pre-diabetes and type 2 diabetes development. Poor dietary habits, characterized by high consumption of processed meats, low-quality carbohydrates, sugary drinks, and low intake of plant-based foods, contribute to chronic inflammation and insulin resistance. The gut microbiome's influence on metabolism and immunity is increasingly recognized, with its composition and function playing a key role in nutrient extraction and metabolite production. The oral microbiome's connection to hyperglycemia is also established, with high glucose levels promoting bacterial growth and inflammation in periodontal tissues. However, microbiome studies predominantly focus on species composition, which has limitations in capturing the genetic heterogeneity and functional diversity within microbial communities. This study aimed to address these limitations by integrating both species-composition and strain-level analyses.

This study employed a six-month, randomized, controlled, single-blind dietary intervention involving 225 pre-diabetic adults. Participants were randomly assigned to either a personalized postprandial glucose-targeting diet (PPT) or a Mediterranean diet (MED) group. 200 participants completed the study. Data collection included self-reported food consumption logs via a smartphone application, continuous glucose monitoring (CGM), anthropometric measurements, and stool, serum, and subgingival plaque samples. The PPT diet was personalized based on a machine-learning algorithm predicting individual postprandial glycemic responses, while the MED diet served as a standard-of-care control. Both diets resulted in significant changes in macronutrient consumption compared to baseline, with the PPT diet showing a more substantial shift in lipid and carbohydrate intake. Microbiome analysis involved metagenomic sequencing of stool and subgingival plaque samples. Species composition, diversity indices, pathway abundance, serum metabolite concentrations (measured using LC-MS), and cytokine levels (measured using Olink PEA) were determined. Statistical analyses, including Wilcoxon paired signed-rank tests and mediation analyses, were used to evaluate the effects of both diets on the microbiome, metabolites, and cytokines, adjusting for baseline characteristics. A pre-trained model was used to assess the association between changes in gut microbiome composition and serum metabolites.

The PPT diet induced significant changes in 19 gut microbial species, 14 gut and one oral microbial pathways, 86 serum metabolites, and 4 cytokines. The MED diet led to significant changes in 5 gut and one oral microbial species, 18 gut microbial pathways, 27 serum metabolites, and 4 cytokines. Mediation analyses revealed that the gut microbiome modulated the effect of both diets on glycemic, metabolic, and immune measurements. Specifically, several gut species, including subtypes of *F. prausnitzii*, mediated the effects of various dietary components on glycemic markers, metabolites, and cytokines. The change in gut microbiome composition explained 12.25% of the variance in serum metabolite changes. Although the gut microbiome showed greater compositional changes, the oral microbiome exhibited higher genetic-level dynamics. The oral microbiome's genetic diversity was negatively associated with environmental richness and positively associated with species prevalence.

This study's findings demonstrate that dietary interventions significantly alter the microbiome, cardiometabolic profile, and immune response in pre-diabetic individuals. The observed changes, including those in previously identified hyperglycemia markers and novel features, highlight the complex interplay between diet, microbiome, and metabolic health. The mediation analyses confirmed the significant role of the gut microbiome in mediating the effects of diet on various physiological parameters. The strong association between gut microbiome composition and serum metabolite changes further underscores the microbiome's importance in metabolic regulation. The greater genetic-level dynamics of the oral microbiome, despite smaller compositional changes, suggests potential for targeting strain-level effects for therapeutic interventions. The study highlights the potential for personalized dietary interventions to influence microbiome composition and function, leading to desired metabolic outcomes.

This study demonstrates the significant impact of dietary interventions on the microbiome, cardiometabolic profile, and immune response in pre-diabetic individuals. Both PPT and MED diets resulted in substantial changes, with the PPT diet showing a more pronounced effect. The study confirmed the microbiome's mediating role in the diet's impact on various health parameters, suggesting personalized dietary interventions tailored to individual microbiome profiles could be a valuable therapeutic strategy for pre-diabetes management. Future research should focus on further elucidating the mechanisms underlying these changes and explore the potential of microbiome-based therapies, such as probiotics and FMTs, in achieving targeted metabolic outcomes.

The study's limitations include the relatively small sample size, the use of self-reported dietary data, and the exploratory nature of some analyses. The potential influence of other factors not measured in the study, such as physical activity and genetics, could not be fully assessed. The reliance on a pre-trained model for assessing microbiome-metabolite associations might have introduced some bias. Further research with larger, more diverse populations and more comprehensive data collection is necessary to validate these findings.