Comparison of gastric mucosa microbiota in metabolically healthy and unhealthy obesity Authors: Bing Yan, Kun Zhang, Zi da Ma, Chengkun Qin Published: 2025 ABSTRACT BackgroundPrevious studies have established a strong link between intestinal microbiota and metabolism. However, no research has examined the relationship between the gastric microbiome and metabolism status. This study characterizes the gastric microbiome in patients with metabolically healthy obesity (MHO) and compares it with that of patients with metabolically unhealthy obesity (MUO).MethodsGastric microbiota composition was analyzed using deep sequencing of the 5-region (5R) 16 S ribosomal RNA (16 S rRNA) gene.ResultsThe gastric microbiota profiles of MHO differed significantly from those of MUO. PICRUSt functional inference revealed distinct microbial metabolic pathways between the groups. PICRUSt analysis identified 30 differentially regulated pathways, with arginine, proline, tryptophan, and phenylalanine metabolism upregulated in MUO. By contrast, DNA repair and recombination proteins, ribosomes, amino acid-related enzymes, and ribosome biogenesis (KEGG Level 3) were upregulated in MHO.ConclusionThis study reveals significant differences in the gastric microbiome between MHO and MUO, highlighting the intricate relationship between microbiota composition and metabolic health. These findings suggest potential microbiome-based therapeutic targets for metabolic disorders.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13098-025-02075-y. Background Obesity is a global health crisis associated with metabolic complications such as dyslipidemia, hypertension, and hyperglycemia—key risk factors for Type 2 diabetes mellitus and cardiovascular disease [1–3]. Obesity can be sub-classified into metabolically healthy obese (MHO) and metabolically unhealthy obese (MUO), based on the presence or absence of metabolic abnormalities [4–7]. Although factors such as visceral fat content, early weight gain, and adipocyte development contribute to these metabolic differences [8, 9], additional mechanisms underlying the favorable metabolic profile of MHO remain underexplored. The gastrointestinal tract hosts trillions of microorganisms that facilitate digestion, metabolism, immune function, and pathogen defense [10–12]. The gut microbiome influences energy balance, gut barrier integrity, and immune responses, and its composition is altered in patients with obesity, linking it to obesity-related diseases such as Type 2 diabetes, cardiovascular disease, and nonalcoholic fatty liver disease [13–16]. While the stomach was historically considered a sterile environment because of its high acidity, advanced DNA sequencing techniques have revealed the presence of a diverse microbial community, including bacteria and fungi, in gastric tissue [17, 18]. Gastric microbiota composition is linked to gastric health, with alterations observed in conditions such as chronic gastritis and gastric cancer [19]. Additionally, the gastric microbiota interacts with microbial communities throughout the gastrointestinal tract, including the oral cavity, duodenum, and small intestine, which have been implicated in inflammatory bowel disease, celiac disease, and colorectal cancer [18, 20]. While extensive research has explored the gut microbiome role in obesity and metabolic disorders, no studies have specifically examined the gastric microbiota composition in MHO and MUO. To address this gap, the present study analyzed the relative abundance of gastric microbial taxa in gastric tissue samples from MHO and MUO undergoing metabolic surgery. Therefore, we aimed to compare the gastric mucosal microbiota between MHO and MUO patients and identify taxa and pathways associated with metabolic health. Methods Patients Stomach tissues from 31 patients who underwent bariatric surgery at Jinan Central Hospital between April 2021 and April 2022 were collected for subsequent 5-region (5R) 16 S ribosomal RNA (16 S rRNA) detection. The study design is shown in Fig. 1. Total weight loss (TWL%) and percent of excess weight loss (EWL%) were calculated based on the results of weight loss obtained after one year of following-up patients. Fig. 1Flow chart showing the experimental procedure Flow chart showing the experimental procedure Definition of MUO and MHO The study classified obesity based on a body mass index (BMI) of ≥ 27.5 kg/m², as the participants were Chinese. They were categorized into two distinct groups: MHO and MUO, depending on the presence of metabolic disorders, such as diabetes, hypertension, or dyslipidemia. Individuals with at least one of these conditions were classified as MUO. The criteria for these metabolic disorders were as follows: (1) waist circumference > 90 cm for males or > 85 cm for females, (2) fasting glucose levels of ≥ 6.1 mmol/L, (3) triglyceride levels of ≥ 1.7 mmol/L, (4) high-density lipoprotein levels < 1.04 mmol/L, (5) blood pressure of ≥ 130/85 mmHg. All measurements were conducted in the morning under fasting conditions, following a 10-minute rest period. Each measurement was performed in duplicate on the non-dominant arm using an automated oscillometric sphygmomanometer (Omron HEM-7136, Omron Healthcare, Kyoto, Japan), the average of the two measurements was used for analysis. Ultimately, 15 individuals classified as MHO and 16 considered MUO were selected for further analysis. None of the participants had taken antibiotics or gastric medications within three months prior to the collection of samples. All gastric tissue samples were obtained from both the MHO and MUO groups through bariatric surgery procedures. These samples were preserved in liquid nitrogen for subsequent analysis. The inclusion criteria required participants to: (1) be between 18 and 65 years; (2) had not been exposed to antibiotics, probiotics, or proton pump inhibitors within one month; (3) have not undergone other gastroenterological procedures within one month; (4) have no history of mental health issues, autoimmune diseases, or psychological disorders. All criteria were verified by detailed medical history obtained from hospital records. Subsequently, the selected participants completed physical evaluations, including height, weight, waist size, and blood pressure measurements. An automatic blood analyzer was utilized to evaluate blood levels of glucose, total cholesterol, triglycerides, low-density lipoprotein, high-density lipoprotein, and uric acid (Cobas 8000, Roche Diagnostics, Basel, Switzerland). DNA extraction, 5R 16 S rRNA gene amplification, and sequencing Sample preparation and sequencing were conducted by Hangzhou Lianchuan Biotechnology Co., Ltd. Briefly, genomic DNA from stomach tissues was extracted using a commercial kit (DP302-02, TIANGEN Biotech, Beijing, China) and subsequently subjected to multiplex amplification of the target regions prior to sequencing. Specific primers were employed to amplify five regions (V2, V3, V5, V6, and V8) of the 16 S ribosomal DNA. (F1, 5′-TGGCGAACGGGTGAGTAA-3′; F2, 5′-ACTCCTACGGGAGGCAGC-3′; F3, 5′-GTGTAGCGGTGRAATGCG-3′; F4, 5′-GGAGCATGTGGWTTAATTCGA-3′; F5, 5′-GGAGGAAGGTGGGGATGAC-33′; R1, 5′-AGACGTG TGCTCTTCCGAAGACGTGTGCTCTTCCGA-3′; R2, 5′-AGACGTGT GCTCTTCCGATCTGTATTACCGCGGCTGCTG-3′; R3, 5′-AGAC GTGTGCTCTTCCGATCTCCCGTCAATTCMTTTGAGTT-3′; R4, 5′-AGACGTGTGCTCTTCCGATCTCGTTGCGGGACTTAACCC-3′; R5,5′-AGACGTGTGCTCTTCCGATCTAAGGCCCGGGAACGTATT-3′; FF1,5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTTGGCGAACGGGTGAGTAA-3′; FF2, 5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTACTCCTACGGGAGGCAGC-3′). The libraries were sequenced on the Illumina NovaSeq6000 system (Illumina San Diego, CA, USA). Statistical analyses Statistical analyses were conducted using SPSS software version 25.0 (SPSS, Chicago, IL, USA). Permutational multivariate analysis of variance (PERMANOVA) was used to assess whether two groups were significantly different. Different controls were included, such as sampling controls, DNA extraction controls and no-template PCR amplification controls, to account for the various sources of contamination from the hospital and laboratory environments as well as the different stages of handling and processing of the samples. To compare continuous variables between groups, t-tests were applied, while Fisher’s exact test was used for categorical variables, which is particularly suitable for small sample sizes. p < 0.05 was deemed statistically significant. The quality filtering was done using parameters applied with VSEARCH and USEARCH: vsearch --fastq_mergepairs raw_R1.fastq.gz --reverse raw_R2.fastq.gz \--fastqout merged.fastq \. --fastqout_notmerged_fwd unmerged_R1.fastq–fastqout_notmerged_rev unmerged_R2.fastq \--fastq_minovlen 20 --fastq_maxdiffs 3 --fastq_minmergelen 380 --fastq_maxmergelen 480 \--threads 8 vsearch --fastq_filter merged.fastq --fastqout filtered.fastq \--fastq_maxee 1.0 --fastq_minlen 380 --fastq_maxlen 480 \ --fastq_truncqual 20 --relabel Sample01. --threads 8. The SMURF package was utilized to aggregate read counts from the five regions into a coherent profiling result, addressing a maximum likelihood problem. The GreenGenes database served as the reference. The assessment of alpha diversity, which measures the diversity within individual samples, was performed using the Mann–Whitney U test. This test provided insights into the variability among the bacterial populations, statistical significance was defined as p < 0.05. Moreover, rarefaction curves were employed to assess if the sample size adequately reflected the total diversity of the bacterial community. The clustering of communities was examined using principal coordinate analysis (PCoA), which is an effective method for identifying patterns and connections within multidimensional datasets. The results of this analysis were further evaluated using the Adonis test, with a significance level once again set at p < 0.05. To identify unique gastric microbiota across varying groups and to explore potential biomarkers that could differentiate these groups, the linear discriminative analysis effect size (LEfSe) approach was employed. We analyzed the LEfSe results using the SILVA database release 138.2. We reported both the linear discriminant analysis (LDA) score ≥ 3 and p < 0.05 as dual criteria to identify taxa with plausible biological relevance. Finally, the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt, version 2.03) was conducted to uncover the various functions of gastric bacteria and to categorize them based on KEGG Orthology functions. The results of the significant findings, with p < 0.05, were effectively illustrated through figures generated by STAMP version 2.1.3, highlighting the crucial outcomes of this research. Result Clinical information A total of thirty-one participants took part in this research. The detailed demographic and clinical data of the participants, categorized into MHO and MUO groups, can be found in Table 1. Out of the 31 individuals, 49.4% (15) were classified as MHO, while 51.6% (16) fell under the MUO category. No significant differences were observed regarding age, height, BMI, pressure, low-density lipoprotein cholesterol and uric acid levels (p > 0.05). However, there were differences in weight, waist, triglycerides, high-density lipoprotein cholesterol and fasting glucose levels in the two groups (p < 0.05). A statistically significant difference (p = 0.016) was observed between the two groups using PERMANOVA. Clinical and follow-up data were compared across both groups, revealing no notable difference in EWL and TWL. Table 1 presents the baseline clinical characteristics of the patients, with 15 samples collected from the MHO group and 16 from the MUO group. Table 1Demographic characteristics of MHO and MUOMHO (n = 15)MUO(n = 16)p valueSex (female), n (%)9, 53%11, 62%0.716Age29.4 ± 8.1333.8 ± 8.560.153Weight (kg)97.9 ± 13.5122.3 ± 26.50.003Height (cm)165.0 ± 6.7170.9 ± 8.50.039BMI (kg/m2)39.4 ± 13.741.6 ± 7.00.577Waist (cm)113.6 ± 16.9131.9 ± 16.90.005Diastolic Blood Pressure (mmHg)82.3 ± 13.388.4 ± 15.10.243Systolic Blood Pressure (mmHg)125.7 ± 7.8137.0 ± 25.90.114Triglycerides (mmol/L)1.3 ± 0.72.0 ± 0.70.004High-density lipoprotein cholesterol (mmol/L)1.2 ± 0.31.0 ± 0.30.042Low-density lipoprotein cholesterol (mmol/L)2.4 ± 0.72.7 ± 0.60.295Uric acid (umol/L)355.3 ± 94.8400.4 ± 129.70.280Fasting Glucose (mmol/L)4.5 ± 0.76.2 ± 0.10.006EWL (%)116 ± 23.898.9 ± 34.50.31TWL (%)38.2 ± 11.536.5 ± 70.47MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Table 1Demographic characteristics of MHO and MUOMHO (n = 15)MUO(n = 16)p valueSex (female), n (%)9, 53%11, 62%0.716Age29.4 ± 8.1333.8 ± 8.560.153Weight (kg)97.9 ± 13.5122.3 ± 26.50.003Height (cm)165.0 ± 6.7170.9 ± 8.50.039BMI (kg/m2)39.4 ± 13.741.6 ± 7.00.577Waist (cm)113.6 ± 16.9131.9 ± 16.90.005Diastolic Blood Pressure (mmHg)82.3 ± 13.388.4 ± 15.10.243Systolic Blood Pressure (mmHg)125.7 ± 7.8137.0 ± 25.90.114Triglycerides (mmol/L)1.3 ± 0.72.0 ± 0.70.004High-density lipoprotein cholesterol (mmol/L)1.2 ± 0.31.0 ± 0.30.042Low-density lipoprotein cholesterol (mmol/L)2.4 ± 0.72.7 ± 0.60.295Uric acid (umol/L)355.3 ± 94.8400.4 ± 129.70.280Fasting Glucose (mmol/L)4.5 ± 0.76.2 ± 0.10.006EWL (%)116 ± 23.898.9 ± 34.50.31TWL (%)38.2 ± 11.536.5 ± 70.47MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Demographic characteristics of MHO and MUO MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Sequencing coverage The investigation into the 5R 16 S rRNA gene was performed on a cohort of 15 patients from the MHO group and 16 patients from the MUO group. We produced a cumulative total of 4,507,588 raw reads (averaging 145,406 reads per sample) and 4,152,895 clean reads (roughly 133,964 reads per sample) across the 31 samples, attaining an average Q30% of 95%. The alpha rarefaction curve indicated sufficient sequencing depth, while Good’s coverage neared 0.99 for both groups. A total of 1804 OTUs were delineated at the 97% sequence-similarity threshold. Bacterial species that exhibited detection rates exceeding 30% in the negative control groups were excluded from the analysis. The removed species are detailed in Supplementary Table S1. Differences in the gastric microbiota diversity Differences in the distribution of gastric microbiota were noted between the two groups examined at the taxonomic levels of phylum, genus, and species. In total, 19 phyla, 351 genera, and 938 species were detected (Figs. 2A, C, E). Analysis of the top 10 taxa at the phylum level (Fig. 2B) revealed that Proteobacteria (62.79%) was the most prevalent in the MUO group, succeeded by Firmicutes (13.35%), Actinobacteria (17.06%), and Bacteroidetes (1.84%). In comparison, the majority of phyla recognized within the MHO group were classified as Proteobacteria (37.68%), with Firmicutes (37.37%) following closely, alongside Actinobacteria (13.62%) and Bacteroidetes (8.57%). At the genus level (Fig. 2D), the MUO group exhibited dominance of Burkholderia (8.68%), Corynebacterium (7.45%), Lactobacillus (6.83%), Pseudomonas (4.57%). Conversely, Lactobacillus (11.79%) represented the highest relative abundance genus in the MHO group. At the species level (Fig. 2F) the most abundant classified reads matched to Burkholderia arboris in the MUO group, while Lactobacillus lactis held the primary share within the MHO group (Fig. 2E). To assess and analyze the richness and diversity of the gastric microbiota, both alpha and beta diversity analyses were conducted. The alpha diversity metrics, which included Chao1, Shannon index, Simpson index, and the count of observed species, were utilized to create four-column diagrams. As depicted in Fig. 3, no significant differences were noted in the Shannon and Simpson indices between the two groups (p > 0.05). Conversely, Chao1 (p = 0.01) and observed species (p = 0.02) indicators in the MHO group were found to be considerably higher than those in the MHO group (p < 0.05). These findings highlight a substantial decrease in gastric microbial richness within the MUO group (Fig. 3). A beta diversity analysis was conducted to examine the variations in the gastric microbiota composition between the two groups. We utilized PCoA to assess the relative abundance of bacteria, while the beta diversity index was calculated with a perm analysis of variance (ANOVA). As shown in Fig. 4, the distribution and composition of gastric microbiota in the MHO group displayed notable differences when compared to the MUO group (p < 0.05) (Fig. 4). Fig. 2Variations in the distribution of bacteria among the MHO and MUO groups. Venn diagram illustrates the overall bacterial counts at the phylum (A), genus (C), and species (E) levels. The bar chart displays the leading 30 taxa of the gastric microbiota for both groups at the phylum (B), genus (D), and species (F) levels. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Variations in the distribution of bacteria among the MHO and MUO groups. Venn diagram illustrates the overall bacterial counts at the phylum (A), genus (C), and species (E) levels. The bar chart displays the leading 30 taxa of the gastric microbiota for both groups at the phylum (B), genus (D), and species (F) levels. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Fig. 3Analysis of alpha diversity in the gastric microbiota was conducted for the MHO and MUO groups. The Mann–Whitney U test was utilized to examine observed species (A), Chao1 (B), Shannon index (C), and Simpson index (D). A statistically significant result was indicated by *p < 0.05. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Analysis of alpha diversity in the gastric microbiota was conducted for the MHO and MUO groups. The Mann–Whitney U test was utilized to examine observed species (A), Chao1 (B), Shannon index (C), and Simpson index (D). A statistically significant result was indicated by *p < 0.05. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Fig. 4Red dots signify the MHO group, while blue dots denote the MUO group. Ovals illustrate the 95% confidence interval for every group. The percentages displayed on the X (PCoA1) and Y (PCoA2) axes reflect the extent of variation. Adonis tests were conducted to assess statistical significance. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Red dots signify the MHO group, while blue dots denote the MUO group. Ovals illustrate the 95% confidence interval for every group. The percentages displayed on the X (PCoA1) and Y (PCoA2) axes reflect the extent of variation. Adonis tests were conducted to assess statistical significance. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Examination of biomarkers We sought to differentiate the microbiota composition between the two groups through the application of the LEfSe algorithm to pinpoint potential biomarkers. Our study revealed notable disparities in the abundance of Alphaproteobacteria, Rhizobiales, Corynebacterium, Sphingomonadaceae and Brevundimonas, which were found to be more prevalent in the MUO cohort. Samples from the MHO group also exhibited elevated levels of Bacilloot, Bacilli, Lactobacillales, Streptococcaceae Negativicutes and Veillonellaceae compared with those obtained from the MUO group (Fig. 5A). Identical results were obtained using the cladogram analysis (Fig. 5B). Fig. 5Bacterial differentiation between MHO and MUO. (A) The LefSe analysis indicates the enrichment of various taxa. A histogram displaying LDA scores (log10) illustrates the bacteria that are more prevalent in the MHO group compared to the MUO group at the phylum, class, order, family, genus, and species levels. Statistically significant differences were noted only when p < 0.05. (B) Cladogram representation. The circular diagram, extending from the center to the outer edge, signifies the classification levels ranging from phylum to genus or species. Yellow indicates species with no statistically significant differences. Each group is represented by specific colors for biomarkers; green nodes denote bacteria that are essential to the MHO group, while red nodes signify microbial categories that are important for the MUO group. Abbreviations for each taxon include: phylum (p_), class (c_), order (o_), family (f_), genus (g_), and species (s_). Statistically significant differences were noted only when p < 0.05. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Bacterial differentiation between MHO and MUO. (A) The LefSe analysis indicates the enrichment of various taxa. A histogram displaying LDA scores (log10) illustrates the bacteria that are more prevalent in the MHO group compared to the MUO group at the phylum, class, order, family, genus, and species levels. Statistically significant differences were noted only when p < 0.05. (B) Cladogram representation. The circular diagram, extending from the center to the outer edge, signifies the classification levels ranging from phylum to genus or species. Yellow indicates species with no statistically significant differences. Each group is represented by specific colors for biomarkers; green nodes denote bacteria that are essential to the MHO group, while red nodes signify microbial categories that are important for the MUO group. Abbreviations for each taxon include: phylum (p_), class (c_), order (o_), family (f_), genus (g_), and species (s_). Statistically significant differences were noted only when p < 0.05. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Predictive functional profiling of gastric microbiota The evaluation of the potential functional profile of gastric microbiota was conducted through PICRUSt functional inference in both groups. As illustrated in Fig. 6, multiple impaired pathways were identified for both groups under investigation. Notably, at KEGG Level 1, four functions attained statistical significance at p < 0.05. Specifically, the MUO group exhibited elevated levels of cellular processes, environmental information processing, and human diseases when compared with the MHO group. Conversely, genetic information processing was found to be heightened in the MHO group in comparison to the MUO group (Fig. 6A, Supplementary Table S2). For Level 2, a comprehensive analysis revealed that a total of 21 functions achieved statistical significance, again at p < 0.05. Individuals categorized as MUO demonstrated an enhancement in various pathways connected to crucial biological processes, such as membrane transport, amino acid metabolism, and carbohydrate metabolism. However, the MHO group demonstrated greater prominence in pathways associated with nucleotide metabolism, cofactor, and vitamin metabolism, as well as pathways pertinent to replication and repair processes (Fig. 6B, Supplementary Table S3). At Level 3, the investigation identified 30 functions that were classified as statistically significant (p < 0.05), which further highlighted the distinctions in the metabolic classifications of the two groups. This analysis highlights the complex and varied functional profiles of gastric microbiota, highlighting the distinct metabolic pathways that characterize the MUO and MHO populations. PICRUSt predicted the MUO group upregulation of multiple pathways, including metabolism of arginine and proline, tryptophan metabolism, and phenylalanine metabolism when compared to the MHO group. Conversely, the MHO group displayed enhanced levels of certain other pathways, which included DNA repair and recombination proteins, ribosomal components, amino acid-related enzymes, and processes of ribosome biogenesis (Fig. 6C, Supplementary Table S4). Fig. 6The PICRUSt functional inference of the gastric microbiota among participants from the two groups. (A) Levels 1, (B) Levels 2, (C) Levels 3 revealed differences in microbiota characterized by distinct functional pathways. Pathways that were downregulated indicated a reduced mean proportion of relative abundance, whereas upregulated pathways pointed to an increased mean proportion of relative abundance. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity The PICRUSt functional inference of the gastric microbiota among participants from the two groups. (A) Levels 1, (B) Levels 2, (C) Levels 3 revealed differences in microbiota characterized by distinct functional pathways. Pathways that were downregulated indicated a reduced mean proportion of relative abundance, whereas upregulated pathways pointed to an increased mean proportion of relative abundance. MHO: metabolically healthy obesity, MUO: metabolically unhealthy obesity Discussion This study categorized patients with obesity into MUO and MHO groups and analyzed their gastric mucosal tissues using 5R sequencing to compare bacterial colonization patterns. Initially, this study characterizes the predominant gastric microbiota in individuals with MHO and MUO. Notably, the present study excluded individuals infected with Helicobacter pylori, whereas earlier investigations frequently enrolled Helicobacter pylori-positive subjects. This design confirms that the previously reported predominance of Helicobacter pylori in gastric microbiota studies is largely attributable to differences in study populations rather than an intrinsic feature of the gastric ecosystem. Previous work has demonstrated that Helicobacter pylori infection contributes to insulin resistance through multiple mechanisms, including elevated pro-inflammatory cytokines, eicosanoids, acute-phase proteins, reactive oxygen species, and altered serum adipocytokine profiles [21]. Moreover, Helicobacter pylori infection is significantly associated with dysregulated triglyceride and HDL levels [22]. Consequently, we deliberately excluded participants infected with Helicobacter pylori, thereby enabling a focused examination of microbiota differences between metabolically distinct phenotypes. This study represents the comprehensive analysis of gastric microbiota distribution in obese patients. There were no notable variations in age or BMI between the two groups. The predominant intragastric flora align with previous reports indicating that Proteobacteria and Firmicutes are the most abundant bacterial phyla. However, their relative frequencies were lower than those reported by Delgado and Gazi [23, 24], potentially owing to differences in dietary habits, geographical origins, or ethnicity. Previous studies have demonstrated that the duodenal microbiota is predominantly composed of these two phyla [25], suggesting similar microbial environments in the upper gastrointestinal tract. In contrast, fecal samples revealed that Firmicutes constituted 77.33%, a proportion significantly higher than that observed in the stomach [26]. This finding underscores the distinct microbial communities present in the upper and lower gastrointestinal tracts. Our results revealed significant differences in gastric microbiota composition between MUO and MHO. The ratio of Firmicutes to Bacteroidetes was found to be greater in the MUO group. Research suggests that a higher Firmicutes/Bacteroidetes ratio correlates with a greater risk of metabolic disorders, inflammation, and diseases related to obesity [27, 28]. Furthermore, a higher ratio has been linked to an increased likelihood of cancer development [29, 30]. Previous studies have reported differences in gut microbiota between MUO and MHO groups. In contrast to our findings, they identified a greater abundance of Alistipes, Bifidobacterium, Eubacterium, Faecalibacterium, and Subdoligranulum in MHO, along with lower levels of Fusobacterium and Megamonas in MUO [31]. These discrepancies may be due to variations in microbial community structures across different gastrointestinal regions [32]. The gastric microbiota of individuals with MUO was dominated by Proteobacteria (62.79%), a phylum associated with inflammation and dysbiosis in metabolic disorders. However, the MHO group exhibited a more balanced distribution of Proteobacteria (37.68%) and Firmicutes (37.37%), with Lactobacillus being the most abundant genus. This finding supports emerging evidence that Lactobacillus species may provide metabolic benefits through anti-inflammatory effects and gut barrier stabilization. Reduced alpha diversity (Chao1 and Observed species) in MUO suggests an unstable microbial ecosystem, potentially contributing to metabolic dysfunction. Beta diversity analysis further confirmed distinct microbial communities, driven by taxa such as Burkholderia and Corynebacterium, which have been linked to insulin resistance and lipid dysregulation. Compared with the MUO group, Veillonella relative abundance was elevated in the MHO group, consistent with previous studies showing lower Veillonella levels in the gut of patients with diabetes, suggesting a potential role in blood glucose regulation [33]. PICRUSt-based functional predictions suggested an enrichment of pathways related to amino acid and carbohydrate metabolism were upregulated in MUO individuals. Both of these pathways are implicated in insulin resistance and chronic inflammation [34–36]. Conversely, MHO individuals exhibited increased relative abundance of pathways involved in DNA repair and ribosome biogenesis, suggesting enhanced cellular maintenance mechanisms that may help mitigate metabolic stress. These findings align with the “healthy microbiome” hypothesis, in which microbial metabolic activity influences host energy homeostasis and oxidative stress responses. Despite observed microbiota differences, follow-up revealed no significant difference in weight loss outcomes between the two groups one year after bariatric surgery. This suggests that bariatric surgery has similar weight - loss effects on both MHO and MUO. However, this is only a short-term effect. We will continue to monitor the patient’s weight loss progress while collecting additional metabolism-related indices to observe the long-term effects of weight loss. Mechanistic studies on bariatric surgery have revealed that the intestinal microbiota undergoes significant changes post-surgery [37]. Future research should focus on investigating the differences in gastric microbiota among obese patients with distinct metabolic phenotypes following bariatric surgery. Our findings have profound implications for clinical practice and policies. Identifying distinct microbial biomarkers could enable the development of innovative diagnostic tools for assessing metabolic health in patients with obesity, thereby advancing personalized treatment strategies. Targeting gastric microbiota could thus become a viable strategy for managing obesity and preventing its associated comorbidities. Integrating microbiota profiling into routine clinical assessments could also help stratify patients based on metabolic risk, improving patient outcomes and reducing healthcare costs [38]. However, this study has limitations. The current sequencing depth and sample size impose constraints on our ability to detect low-abundance taxa and subtle differences, potentially leading to the oversight of rare microbial signals. In our future studies, we will expand the sample size and increase multi-center samples to construct a multivariable model. Consequently, validation in larger, multicenter cohorts is essential to confirm these candidate biomarkers and enhance reproducibility. Furthermore, the functional profiles derived from PICRUSt are predictive inferences based solely on 16 S rRNA marker-gene sequences, they do not accurately reflect actual gene relative abundance, enzymatic activity, or metabolite concentrations. Therefore, causal interpretations must await experimental verification through metatranscriptomics, metaproteomics, or metabolomics. Moreover, ethical concerns over repeated gastric biopsy collection prevented analysis of postoperative microbiota changes. The absence of longitudinal data also limits the ability to assess long-term microbiota dynamics and metabolic health. While our findings highlight microbial differences between MHO and MUO groups, clinical validation of these biomarkers is necessary for their application in medical practice. Future animal studies will be crucial in elucidating the relationship between gastric microbiota alterations and obesity-related metabolic complications. Conclusion Overall, this study provides insights into the distinct gastric microbiota compositions and clinical characteristics of MHO and MUO phenotypes. These findings underscore the potential of microbial biomarkers in obesity assessment and management, paving the way for microbiome-targeted therapeutic strategies to improve patient outcomes. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3. Supplementary Material 4. Supplementary Material 4.