Dynamics of Gut Microbiota and Clinical Variables after Ketogenic and Mediterranean Diets in Drug-Naïve Patients with Type 2 Diabetes Mellitus and Obesity

Abstract

Type 2 diabetes mellitus (T2DM), the most common form of diabetes, is a progressive chronic metabolic disease that has increasingly spread worldwide, enhancing the mortality rate, particularly from cardiovascular diseases (CVD). Lifestyle improvement through diet and physical activity is, together with drug treatment, the cornerstone of T2DM management. The Mediterranean diet (MD), which favors a prevalence of unprocessed vegetable foods and a reduction in red meats and industrial foods, without excluding any food category, is usually recommended. Recently, scientific societies have promoted a very low-calorie ketogenic diet (VLCKD), a multiphasic protocol that limits carbohydrates and then gradually re-introduces them, with a favorable outcome on body weight and metabolic parameters. Indeed, gut microbiota (GM) modifications have been linked to overweight/obesity and metabolic alterations typical of T2DM. Diet is known to affect GM largely, but only a few studies have investigated the effects of VLCKD on GM, especially in T2DM. In this study, we have compared anthropometric, biochemical, lifestyle parameters, the quality of life, and the GM of eleven patients with recently diagnosed T2DM and overweight or obesity, randomly assigned to two groups of six and five patients who followed the VLCKD (KETO) or hypocaloric MD (MEDI) respectively; parameters were recorded at baseline (T0) and after two (T2) and three months (T3). The results showed that VLCKD had more significant beneficial effects than MD on anthropometric parameters, while biochemical improvements did not statistically differ. As for the GM, despite the lack of significant results regarding the alpha and beta diversity, and the Firmicutes/Bacteroidota ratio between the two groups, in the KETO group, a significant increase in beneficial microbial taxa such as Verrucomicrobiota phylum with its members Verrucomicrobiae, Verrucomicrobiales, Akkermansiaceae, and Akkermansia, Christensenellaceae family, Eubacterium spp., and a reduction in microbial taxa previously associated with obesity (Firmicutes and Actinobacteriota) or other diseases (Alistipes) was observed both at T2 and T3. With regards to the MEDI group, variations were limited to a significant increase in Actinobacteroidota phylum at T2 and T3 and Firmicutes phylum at T3. Moreover, a metagenomic alteration linked to some metabolic pathways was found exclusively in the KETO group. In conclusion, both dietary approaches allowed patients to improve their state of health, but VLCKD has shown better results on body composition as well as on GM profile.

Keywords: 16S rRNA; Mediterranean diet; body composition; glycometabolic status; gut microbiota; ketogenic diet; obesity; physical activity; quality of life; type 2 diabetes mellitus.

Figure 1

Amplicon sequence variants (ASVs) rarefaction curves per sample. Reads were selected by random subsampling without replacement at incremental steps of 50 reads.

Figure 2

Gut microbiota alpha diversity comparison between diets at each time point. Each subplot concerns a different alpha diversity metric (Observed ASVs, Shannon, or Pielou’s J). The boxes highlight the median value and cover the 25th and 75th percentiles, with whiskers extending to the more extreme value within 1.5 times the length of the box. Statistical significance was evaluated by the Mann–Whitney U test, and it was indicated as follows: ns, non-significant; *, p ≤ 0.05. KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD), MEDI = 5 patients who followed a low-calorie Mediterranean diet (MD).

Figure 3

Gut microbiota alpha diversity comparison between time points for each diet. Each subplot concerns a different combination of alpha diversity metric (Observed ASVs, Shannon, or Pielou’s J) and diet (KETO or MEDI). Samples from the same patient are linked by a gray line. The colored lines depict the mean values at each timepoint. Statistical significance was evaluated by the paired Wilcoxon signed-rank test, and it was indicated as follows: ns, non-significant. KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD), MEDI = 5 patients who followed a low-calorie Mediterranean diet (MD). Samples were analyzed at baseline (T0), after two months (T2), and after three months (T3) of nutritional intervention.

Figure 4

Principal coordinate analysis plots showing gut microbiota compositional changes along time points for each diet. Each subplot concerns a different beta diversity metric (Bray-Curtis, unweighted UniFrac, or weighted UniFrac). Samples from the same patient are linked by a gray line. KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD), MEDI = 5 patients who followed a low-calorie Mediterranean diet (MD). Samples were analyzed at baseline (T0), after two months (T2), and after three months (T3) of nutritional intervention.

Figure 5

Principal coordinate analysis plots with comparisons of gut microbiota composition between diets at each time point. Each subplot concerns a combination of beta diversity metric (Bray-Curtis, unweighted UniFrac, or weighted UniFrac) and timepoint (T0, T2, or T3). Ellipsoids depict the 90% compositional confidence interval. Statistical significance was evaluated by the PERMANOVA test, with statistical summaries included in each subplot. KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD), MEDI = 5 patients who followed a low-calorie Mediterranean diet (MD).

Figure 6

Principal coordinate analysis plots with comparisons of gut microbiota composition between time points for each diet. Each subplot concerns a combination of beta diversity metric (Bray-Curtis, unweighted UniFrac, or weighted UniFrac) and diet (KETO or MEDI). Ellipsoids depict the 90% compositional confidence interval. Samples from the same patient are linked by a gray line. Statistical significance was evaluated by the PERMANOVA test, with statistical summaries included in each subplot. KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD), MEDI = 5 patients who followed a low-calorie Mediterranean diet (MD). Samples were analyzed at baseline (T0), after two months (T2), and after three months (T3) of nutritional intervention.

Figure 7

Gut microbiota Firmicutes/Bacteoroidota ratio comparison between time points for each diet. Each subplot concerns a different diet (KETO or MEDI). Samples from the same patient are linked by a gray line. The colored lines depict the mean values at each time point. A log₁₀ y-axis was used. Statistical significance was evaluated by the paired Wilcoxon signed-rank test, and it was indicated as follows: ns, non-significant. KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD), MEDI = 5 patients who followed a low-calorie Mediterranean diet (MD). Samples were analyzed at baseline (T0), after two months (T2), and after three months (T3) of nutritional intervention.

Figure 8

Changes in gut microbiota taxa abundances between time points for each diet. Each subplot concerns a comparison between timepoints (T0 vs. T2, T0 vs. T3, or T2 vs. T3) in one of the diets (KETO, green MEDI, purple). Statistical significance was evaluated by running a Generalized Linear Mixed-effects Model with MaAsLin2. Effect size is represented by the MaAsLin2 model coefficients and respective standard errors. Only taxa abundance changes at p ≤ 0.05 and q ≤ 0.25 are considered statistically significant. q: p adjusted for Benjamini–Hochberg (BH) correction test with a cut-off at q ≤ 0.25. KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD), MEDI = 5 patients who followed a low-calorie Mediterranean diet (MD). Samples were analyzed at baseline (T0), after two months (T2), and after three months (T3) of nutritional intervention.

Figure 9

Relative abundance changes in gut microbiota taxa between time points for each diet. Only taxa significantly enriched or depleted during the nutritional intervention (according to MaAsLin2 models) are shown (at the genus level for KETO and at the phylum level for MEDI). Genera are colored in the KETO plot based on the phylum to which each genus belongs (red: Verrucomicrobia; blue: Firmicutes; green: Bacteroidota). KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD), MEDI = 5 patients who followed a low-calorie Mediterranean diet (MD). Samples were analyzed at baseline (T0), after two months (T2), and after three months (T3) of nutritional intervention.

Figure 10

Changes in gut microbiota predicted function abundances between time points for each diet. Each subplot concerns a comparison between time points (T0 vs. T2, T0 vs. T3, or T2 vs. T3). Statistical significance was evaluated by running a Generalized Linear Mixed-effects Model with MaAsLin2. Effect size is represented by the MaAsLin2 model coefficients and respective standard errors. Only predicted function abundance changes at p ≤ 0.05 and q ≤ 0.25 are considered statistically significant. KETO = 6 patients who followed a very-low-calorie ketogenic diet (VLCKD). Samples were analyzed at baseline (T0), after two months (T2), and after three months (T3) of nutritional intervention.