Clinical results before and after sleeve gastrectomy

We collected gastric microbiome profiling and clinical follow-up data for 12 months after sleeve gastrectomy from 15 individuals without diabetes and H. pylori infection (ND), 6 individuals without diabetes and with H. pylori infection (ND-HP), 11 individuals with diabetes (DM), and 8 individuals with diabetes and H. pylori infection (DM-HP). The majority of participants were in their 30s to 50s. The ND group was significantly younger compared to other groups. Preoperative BMI and blood pressure showed no significant differences, and C-peptide levels were also similar among 4 groups. In the comparison between the 2 groups with DM, the median duration of DM was higher in the DM-Hp group, but this difference was not significant (5.0 [4.5;18.5] in the DM group vs. 9.0 [4.5;11.9] in the DM-Hp group, p = 0.868). The proportion of patients using insulin was higher in the DM group, but this difference was also not statistically significant (72.7% vs. 37.5%, p = 0.181) (Figs. 1, S1 and Tables 1, S2).

Fig. 1: Comparison of clinical changes by group before (pre-) and after (12M) sleeve gastrectomy.

Each group (ND, ND-HP, DM, and DM-HP) was compared using Wilcoxon and t tests for several clinical parameters, including A aspartate transferase (AST), B hemoglobin A1c (HbA1c), C high-density lipoprotein (HDL), D serum uric acid, E body mass index (BMI), F blood glucose and G hepatic steatosis index (HSI). ND non-DM without H. pylori infection, ND-HP non-DM with H. pylori infection, DM diabetes mellitus, DM-HP DM with H. pylori infection.

Table 1 Baseline characteristics for this study.

The weight loss at 12 months postoperatively (TWL and EWL) did not show significant differences among 4 groups, but the ND group had the highest median TWL (31.4%) and the DM-Hp group had the lowest (22.7%). Similarly, the DM-Hp group had the lowest EWL (64.1%). Indicators related to blood glucose, lipids, and hepatic steatosis improved in all groups postoperatively (Tables 2 and S2). Especially, significant differences were noted from preoperative to postoperative values in the DM group. The DM group showed statistically significant reductions in fasting glucose, HbA1c, ALT, and the hepatic steatosis index at 12 months postoperatively, and a significant increase in HDL cholesterol. However, the DM-Hp group did not show significant differences in any of these indicators.

Table 2 Postoperative changes of body weight and laboratory findings.

These results suggest that clinical improvement after sleeve gastrectomy is effective in patients with obesity and diabetes, but may be hindered in patients with H. pylori infection.

Composition and diversity of gastric microbiome in patients before sleeve gastrectomy

The relative abundance of the gastric microbiome was compared among the four groups. Firmicutes accounted for 35.6%, 17.5%, 33%, and 18% in the ND, ND-HP, DM, and DM-HP groups, respectively. Actinobacteriota were dominant in the H. pylori-negative groups, accounting for 44.2% and 46.9% in the ND and DM groups, respectively. Actinobacteriota represented 25.5% and 28.7% in the ND-HP and DM-HP groups, respectively. Notably, Campilobacterota were dominant in the ND-HP and DM-HP groups, accounting for 46.4% and 34.8%, respectively (Fig. S3). Cutibacterium was the most dominant genus in the ND and DM groups, accounting for 38.11% and 42.8%, respectively. As expected, in the ND-HP and DM-HP groups, H. pylori was the most dominant species, accounting for 45.97% and 33.96%, respectively. These results indicated that Firmicutes were prevalent in all groups; however, their relative abundance varied. Actinobacteriota were more common in the H. pylori-negative groups, whereas Campilobacterota were more common in the H. pylori-positive groups. The genus Cutibacterium was predominant in the H. pylori-negative group, whereas Helicobacter was predominant in the H. pylori-positive group (Fig. 2A).

Fig. 2: Relative abundance and diversity of gastric microbiome in the ND, ND-HP, DM, and DM-HP groups.

A Relative abundance in the ND, ND-HP, DM, and DM-HP groups. B Observed operational taxonomic units (OTUs) (number of distinct features): calculates the number of distinct OTUs. Chao1 index: estimates diversity from abundant data. Pielou’s evenness: measures relative evenness of species richness. Simpson’s index: measures the relative abundance of the different species making up the sample richness. Shannon’s index: accounts for the abundance and evenness of the taxa. C β-diversity plot between the four groups (upper panel). The table shows significant differences through the permutational multivariate analysis of variance test (bottom panel).

The α-diversity decreased in the order of ND > DM > ND-HP > DM-HP (Fig. 2B and Tables S3 and S4). The α-diversity of the DM-HP group was significantly reduced compared to that of the ND group, based on Shannon’s index, Simpson’s index, and Pielou’s evenness (p < 0.01). Moreover, the ND-HP group showed a more significant reduction in diversity than the ND group (p < 0.01). Pielou’s evenness was significantly lower in the ND-HP group than in the DM group (p = 0.02). Significant group differences were observed in α- and β-diversity when comparing the ND and DM-HP and DM and ND-HP groups (p ≤ 0.05) based on H. pylori infection (Fig. 2C and Table S5). Therefore, β-diversity differs depending on the presence of diabetes or H. pylori infection.

Alteration of gastric microbiota in patients with obesity based on the presence of DM and H. pylori

Figure 3 (Fig. S2 and Table S6) shows the results of the network analysis, involving BNC and CC. The plot indicates that in the ND group, Lachnospiraceae-NK4A136-group (0.056) and Lactobacillus (0.051) had high BNC values, whereas Porphyromonas (1), Prevotella (0.86), and Lachnoclostridium (0.809) had high CC values (Fig. 3A). In the ND-HP group, Helicobacter (0.011) and Streptococcus (0.011) showed high BNC values. Staphylococcus (0.90) had a high CC value, whereas Lactobacillus (0.80) had a low CC value (Fig. 3B). In the DM group, Cutibacterium (0.052) had the highest BNC value, and Lachnoanaerobaculum (0.73) had the highest CC value (Fig. 3C). In the DM-HP group, Helicobacter (0.051) had the highest BNC value, and Romboutsia (0.001) had the lowest BNC value. Acinetobacter (0.73) had the highest CC value, and Leptotrichia (0.461) had the lowest (Fig. 3D). In the ND-HP group, the CC value was much higher than 0.40 (horizontal red dotted line: arbitrary boundary line) compared to that of other groups. Therefore, the connectivity between nodes is higher in the ND-HP group than in other groups. These results suggest that the presence of Helicobacter is associated with a higher CC and greater co-occurrence among bacterial genera in gastric microbiota.

Fig. 3: Co-occurrence analysis between the gastric microbiome before sleeve gastrectomy in each group.

The y-axis indicates betweenness centrality (BNC), and the x-axis indicates the clustering coefficient (CC). The genera located at the bottom right of the plots have the lowest rates of co-occurrence with other genera (low CC) and play a central role in the topology of the network (high BNC). The red dotted line is an arbitrary line for comparing the positions above the microbiota on the coordinates of the BNC and CC. The red dotted line on the vertical axis represents a random cutoff based on a BNC of 0.10, and the red dotted line on the horizontal axis represents a random cutoff based on a CC of 0.4. A ND is non-DM without H. pylori infection group, B ND-Hp is non-DM with H. pylori infection group, C DM is diabetes mellitus group, (C) DM-Hp is DM with Indicates H. pylori infection group..

We identified pathways with an LDA score of >3.0 based on the microbiome in each group (Fig. 4A and Table S7). In the ND-HP group, several pathways related to cell motility (LDA = 3.8), bacterial chemotaxis (LDA = 3.76), lipopolysaccharide biosynthesis (LDA = 3.79), and biotin metabolism (LDA = 3.54) showed high LDA scores, indicating that these pathways were more abundant in this group than in the other groups. In the ND group, all features corresponded to pathways in the metabolism category, with a particularly high LDA score for the xenobiotic biodegradation and metabolism pathways. In the DM-HP group, the taurine and hypotaurine metabolism pathways (LDA = 3.2) had high LDA scores, whereas most of the features in the DM group corresponded to pathways in the metabolism category.

Fig. 4: Comparison of functions using PICRUSt2 from the microbiome in each group.

A Results of LEfSe analysis: the y-axis represents a gastric microbial-related pathway, and the x-axis indicates the linear discriminant analysis (LDA) score. B Heatmap of the relative abundance of the microbiome according to features based on the LDA in each group.

The microbiome abundance patterns in the ND-HP and DM-HP and ND and DM pairs were similar based on the heatmap results (Fig. 4B and Table S8), indicating that the pathways corresponding to these pairs were associated with certain characteristics depending on the presence or absence of Helicobacter. Therefore, when considering the characteristics of the strains representing each group, these pathways can be used to differentiate between the groups based on the presence or absence of Helicobacter.

Correlation of clinical parameters and the gastric microbiome before sleeve gastrectomy

We performed a correlation analysis between the gastric microbiome and clinical indicators to confirm the preoperative gastric environment and clinical relationships in each group. Figure 5 shows the correlation analysis between clinical parameters and the predominant top 20 microbial communities in the gastric tissue before sleeve gastrectomy. Notably, the ND-HP group showed a higher correlation with the clinical parameters of the microbiome than the other groups (Fig. 5 and Table S9). The ND group showed a lower correlation between microbes and clinical parameters than the other groups. However, Streptococcus showed a correlation with HbA1c (R = 0.57) and low-density lipoprotein (LDL) (R = 0.584). Bosea was positively correlated with TG (R = 0.547), HDL (R = 0.522), and blood glucose levels (R = 0.587). Fusobacterium (R = 0.535), Tannerella (R = 0.653), and Bosea (R = 0.587) showed strong positive correlations with blood glucose levels, whereas Neisseria (R = −0.635) and Actinomyces (R = −0.571) showed a strong negative correlation. In addition, Bacteria; unclassification positively correlated with HIS (R = 0.621).

Fig. 5: Correlation between the top 20 microbiome and clinical parameters in gastric tissues before sleeve gastrectomy for each group.

Correlation analysis results using Spearman’s analysis. Red indicates a positive correlation, and blue indicates a negative correlation. Asterisks (*) represent the p value of the comparative statistical test for each bacterial genus (*p < 0.05, **p < 0.01, ***p < 0.001).

In the ND-HP group, LDL showed an overall positive correlation, and higher correlations were observed for Staphylococcus (R = 0.828) and Veillonella (R = 0.828). In addition, most clinical parameters showed negative correlations in Rothia, whereas TG showed a relatively strong negative correlation (R = −0.942). Bifidobacterium showed relatively strong positive correlations with ALT (R = 0.811), TG (R = 0.811), and HSI (R = 0.840) levels. The ND-HP group showed negative correlations with blood glucose, insulin, and homeostatic model assessment of insulin resistance (HOMA-IR), except for Helicobacter and Rothia, which were among the top 20 microbiome. In particular, insulin levels had a relatively high overall negative correlation.

In the DM group, negative correlations were observed between Cutibacterium and ALT (R = −0.636) and aspartate transferase (AST) (R = −0.665). Staphylococcus was negatively correlated with HDL levels (R = −0.675). Tannerella and Bacteroides showed a strong positive correlation with ALT (R = 0.769 and R = 0.837, respectively) and AST (R = 0.739 and R = 0.811, respectively). Actinomyces showed strong positive correlations with HbA1c (R = 0.668) and HDL levels (R = 0.539). Bacteria; unclassification showed positive correlations with HDL (R = 0.611).

The DM-HP group showed a negative correlation with Cutibacterium and HDL (R = −0.802), and Helicobacter showed a positive and negative correlation with HDL (R = 0.718) and TG (R = 0.833), respectively. Cutibacterium and Helicobacter showed an inverse correlation with TG and HDL levels.

Cutibacterium and f_Lachnospiraceae; unclassification showed a strong positive correlation with insulin (R = 0.785) and HOMA-IR (R = 0.880). d__Bacteria;_ unclassification showed a negative correlation with HbA1c (R = −0.714) levels. Staphylococcus showed a negative correlation with BMI (R = −0.761). Additionally, Lactobacillus and Rhotia positively correlated with AST (R = 0.761) and ALT (R = 0.913) levels.