Obesity is a widespread metabolic disorder characterized by chronic and complex traits. Its etiology is multifactorial, encompassing genetic, environmental, and behavioral influences, and its prevalence is on the rise globally (Janić et al., 2025). This condition serves as a significant risk factor for various health issues, including cardiovascular disease, type 2 diabetes, non-alcoholic fatty liver disease, and certain cancers (Alexander et al., 2019; Wang et al., 2024; Yu et al., 2024; Leitzmann et al., 2025; Mo et al., 2025). The World Obesity Federation (2025) report projects that the number of adults with elevated body mass index (BMI) worldwide will increase to 2.9 billion by 2030. Similarly, China is experiencing a parallel trend, with the rates of obesity and overweight individuals expected to reach 70.5% by 2030 (World Obesity Federation, 2025). Furthermore, there is a concerning annual increase in the number of overweight adolescents and children, with a noticeable trend towards younger populations becoming obese [NCD Risk Factor Collaboration (NCD-RisC), 2024]. This situation poses a substantial burden on global public health systems. Consequently, it is of paramount clinical and social importance to gain a comprehensive understanding of the pathogenesis of obesity and to identify new targets for early intervention.

In recent years, advancements in microbiome research and high-throughput sequencing technologies have positioned gut microbiota as a significant factor influencing host metabolism and immune function, thereby emerging as a prominent subject of investigation in the field of metabolic disease research. The gut microbiota is essential for modulating energy intake and expenditure, regulating metabolic processes and inflammatory responses, and preserving the integrity of the intestinal barrier (Carmody and Bisanz, 2023; Wu et al., 2024a). Research indicates that germ-free mice subjected to fecal microbiota transplantation for the purpose of addressing obesity exhibited a marked increase in body weight and associated obesity symptoms, thereby implying a potential causal relationship between gut microbiota and obesity (Borrello et al., 2022; Yarahmadi et al., 2024). Furthermore, investigations have revealed that individuals with obesity often present with dysbiosis in their gut microbiota, accompanied by alterations in the abundance of critical bacterial genera. Recent extensive metagenomic analyses have identified Megamonas as a potential “obesogenic bacterium” among obese individuals in China, and its enrichment is significantly positively correlated with BMI and waist circumference (Wu et al., 2024b). Similarly, Enterobacter cloacae has been identified as a significant contributor to obesity. Research indicates that this bacterium constitutes approximately one-third of the total microbial population in the intestines of individuals with obesity. Furthermore, studies have demonstrated that Enterobacter cloacae is capable of inducing obesity and exacerbating inflammation in murine models (Fei and Zhao, 2013). Individuals with obesity frequently exhibit a reduction in the presence of anti-obesity bacteria, including species such as Bacteroides thetaiotaomicron and Faecalibacterium prausnitzii (Liu et al., 2017; Song et al., 2023). And, research indicates that the abundance of Faecalibacterium prausnitzii is significantly and negatively correlated with both weight loss and the reduction of fat mass (Kahleova et al., 2020); the prevalence of Bacteroides thetaiotaomicron demonstrates a negative correlation with serum glutamate levels, which are associated with fat accumulation and weight gain (Liu et al., 2017). Furthermore, an increasing body of research has demonstrated that the intestinal microbiota engage in interactions with the host through the production of metabolites (Nicholson et al., 2012). Microorganisms are capable of generating essential metabolites, including short-chain fatty acids (SCFAs) and bile acid derivatives, through the fermentation of dietary fiber (Delzenne et al., 2025). SCFAs, particularly propionic acid and butyric acid, play a significant role in lipid metabolism by modulating hypothalamic activity via the blood–brain barrier, thereby decreasing food intake and inhibiting lipogenesis (Mukhopadhya and Louis, 2025). Likewise, the synthesis of secondary bile acids has been associated with enhanced energy expenditure and a reduction in fat accumulation (Ticho et al., 2019). In general, current research has established that individuals with obesity frequently exhibit specific traits, including dysbiosis in microbial diversity, a diminished presence of butyrate-producing bacteria (notably Faecalibacterium) and bacteria responsible for producing secondary bile acids, alongside an elevated abundance of pro-inflammatory bacteria. This phenomenon is referred to as the “obesity-type microecological imbalance” (Breton et al., 2022)^.

Existing research has established a significant relationship between obesity and intestinal microecology; however, there is considerable variability in the microecological profiles among individuals with obesity. Initial investigations highlighted an enrichment of Firmicutes in obese populations, accompanied by a reduction in Bacteroidetes, leading to the conclusion that an elevated Firmicutes/Bacteroidetes ratio is a characteristic feature of obesity (Breton et al., 2022). Recent research has introduced alternative viewpoints, with certain findings indicating an elevation in Bacteroidetes among individuals with obesity, or proposing that there is no significant variation in the Firmicutes/Bacteroidetes ratio (Magne et al., 2020). Correspondingly, a prevailing agreement among various studies indicates that individuals with obesity tend to demonstrate a reduced diversity of intestinal microbiota. However, research conducted by Japanese scholars has demonstrated that the diversity of intestinal microbiota in individuals with obesity is significantly higher than that found in individuals without obesity (Uema et al., 2022). These divergent findings highlight the variability in research results concerning intestinal microorganisms among various populations of individuals with obesity.

The alterations and composition of intestinal microbiota are influenced by a variety of factors, including demographic variations (such as ethnicity and geographic location), dietary patterns, lifestyle choices, sample size, and the methodologies employed in analysis (Brooks et al., 2018; Dwiyanto et al., 2021; Borrello et al., 2022; Andreu-Sánchez et al., 2025). More importantly, a study has demonstrated notable variations in the diversity and composition of gut microbiota among individuals with differing BMIs (Wang et al., 2025). Nonetheless, there remain significant gaps in the existing systematic research concerning the relationship between BMI and intestinal microbiota in China. Notably, studies that incorporate normal, overweight, and obese populations within a cohesive framework for comparative analysis are particularly limited. Furthermore, there is a deficiency in the investigation of microbial markers that could facilitate early intervention during the overweight phase. It is also important to acknowledge that BMI does not provide a direct measure of adipose tissue. Individuals with comparable BMI values may exhibit divergent health outcomes, indicating the necessity for additional metrics to assess the health implications associated with overweight and obesity (Sweatt et al., 2024). Consequently, to address these gaps, it is essential not only to undertake systematic research across BMI strata but also to explore the potential of gut microbiota, combined with routine metabolic markers, for stratifying metabolic health independent of BMI itself.

Our study was designed with two primary innovative aims: first, to systematically identify the unique and transitional gut microbial signatures of the overweight state, a critical yet understudied pre-obesity phase; and second, to develop and evaluate a predictive model that integrates these microbial markers with clinical metabolic indicators for obesity stratification, deliberately excluding BMI to test its intrinsic utility. By identifying core intestinal microorganisms associated with different BMI categories and clinical metabolic indicators independent of BMI, we developed a predictive model that integrates microbial markers significantly correlated with both BMI and clinical indicators. This study offers theoretical insights and identifies potential biomarkers for the early detection and targeted intervention of obesity, highlighting the variations in intestinal microorganisms among individuals with differing BMI classifications.

Over the past decade, the gut microbiota has been increasingly recognized as a key contributor to obesity pathogenesis. Substantial evidence demonstrates that obese individuals exhibit significant alterations in microbial composition, functional capacity, and metabolic output, particularly characterized by reduced butyrate-producing bacteria, elevated pro-inflammatory taxa, and impaired SCFAs biosynthesis (Bouter et al., 2017; Sun et al., 2018; Aron-Wisnewsky et al., 2021). However, as research cohorts expand and analytical techniques advance, the conventional definition of an “obesity-associated microbiome” requires critical reevaluation (Magne et al., 2020). Critically, most prior studies have treated obesity as a single state (e.g., obesity vs. normal) or focused exclusively on extreme BMI groups, overlooking the transitional “overweight” phase. From a public health perspective, this intermediate stage represents both a precursor to obesity and the most reversible window for microbial ecological interventions. Our study provides novel insights into the gut microbiota’s role in obesity by focusing on the often-overlooked overweight population and by constructing a diagnostic framework that deliberately excludes BMI.

Our results demonstrate that Ob exhibits a significantly lower Shannon index compared to Nor, reflecting reduced α-diversity of gut microbiota—a finding consistent with prior research (Hu et al., 2022). It is noteworthy that the absolute values of the Shannon index in our cohort (primarily ranging between 2 and 3.7 for the Nor group) appear lower than some published benchmarks. This discrepancy may be attributed to methodological variations (e.g., DNA extraction kits, 16S rRNA gene region targeted [V3-V4], sequencing platform, and bioinformatic pipelines), population-specific factors (e.g., genetic background, geography, and diet of our Chinese cohort), or a combination thereof. While cross-study comparisons of absolute diversity values should be interpreted with caution, the significant relative difference observed between the Ob and Nor groups within our standardized analytical framework remains valid and informative. Notably, no significant α-diversity difference was observed between the Ow and Ob groups. β-diversity analysis revealed a gradient distribution of microbial communities across BMI categories, with Ob and Nor groups forming distinct clusters and Ow positioned intermediately, suggesting BMI as a key driver of microbiota restructuring (Wang et al., 2025).

At the genus level, Coprococcus, Faecalibacterium, Lachnospiraceae UCG-001, and Lachnospiraceae NK4A136 group were significantly enriched in the Nor group. Notably, Faecalibacterium-a genus consistently linked to metabolic health-exhibits reduced abundance in obese individuals, as demonstrated by prior studies (Maioli et al., 2021; Islam et al., 2022). The Lachnospiraceae NK4A136 group and Lachnospiraceae UCG-001 exhibited a positive correlation with SCFA production capacity, a metabolic pathway critically linked to enhanced energy metabolism and attenuated inflammatory responses (Ma et al., 2020). Furthermore, our findings corroborate prior evidence indicating an inverse association between the Lachnospiraceae NK4A136 group and obesity (Companys et al., 2021). Notably, while certain studies report increased Coprococcus abundance in obese individuals, significant functional heterogeneity exists within this genus. Multiple independent cohort and intervention studies demonstrate that specific species (e.g., C. eutactus) correlate with enhanced insulin sensitivity, reduced BMI, and possess butyrate-producing and anti-inflammatory capabilities (Yang et al., 2023; González et al., 2025). Thus, the Coprococcus enrichment observed in our normal-weight group likely represents a “beneficial subpopulation,” with its depletion potentially indicating loss of functionally protective microbiota in obesity. However, the Ruminococcus gnavus group, Morganella, and Clostridium innocuum group were significantly enriched in the Ow cohort. Notably, both the Ruminococcus gnavus group and Morganella are associated with pro-obesity effects (Grahnemo et al., 2022; Sun et al., 2023), whereas the Clostridium innocuum group-a recognized probiotic-exhibits intestinal regulatory and potential anti-obesity properties (Bian et al., 2022; Mei et al., 2022). These findings suggest that the gut microbiota in the Ow stage may represent a transitional equilibrium, characterized by the concurrent expansion of pro-inflammatory taxa and compensatory proliferation of protective flora. The Escherichia-Shigella, Collinsella, and Parabacteroides genera were significantly enriched in the Ob group. Escherichia-Shigella exhibited obesogenic and pro-inflammatory properties in the gut of obese individuals, showing significant positive correlations with obesity markers including BMI (Hong et al., 2024; Gao et al., 2025). Similarly, Collinsella demonstrated consistent positive associations with obesity-related parameters such as TG, TC, and BMI (Gallardo-Becerra et al., 2020; Wang et al., 2020). While Parabacteroides is generally regarded as a potentially beneficial genus, certain studies have reported its positive association with BMI (Chiu et al., 2014; Guan et al., 2021). Notably, P. merdae and P. distasonis demonstrate significant enrichment in individuals with hypertension and diabetes, with Parabacteroides exhibiting particularly high abundance in obese patients with type 2 diabetes. These findings suggest that specific Parabacteroides species may contribute to metabolic dysregulation (Yan et al., 2017; Hasain et al., 2020; Yang et al., 2021). The Parabacteroides genus demonstrates significantly higher abundance in individuals with inflammatory bowel disease or elevated inflammatory indices, suggesting a potential association between Parabacteroides proliferation and systemic inflammation levels (Aranaz et al., 2021; Iliodromiti et al., 2023). This observation aligns with the elevated WBC counts consistently reported in obese populations, which further corroborates systemic inflammatory responses. Notably, our findings reveal increased Parabacteroides abundance in obese individuals, potentially indicating the expansion of specific Parabacteroides subspecies linked to metabolic dysregulation and pro-inflammatory alterations in the gut microbiota. Of particular note is that the abundance of Lactobacillus was significantly increased in the obese group, a result that seems to contradict the “beneficial” image of this genus that is often widely used in probiotic preparations. However, in fact, the Lactobacillus genus is highly heterogeneous, and different species and even different strains of the same species have significant differences in their effects on host metabolism. For example, Lactobacillus acidophilus has been shown to be associated with weight gain, while Lactobacillus plantarum has been shown to improve obesity (Million et al., 2012). Furthermore, a study found that the number of Lactobacillus species in obese patients was significantly higher than lean controls, a finding consistent with our results (Armougom et al., 2009). However, the limitations of 16S rRNA sequencing preclude precise species-level differentiation. We therefore hypothesize that the enriched Lactobacillus population in obesity may represent an obesity-promoting subpopulation or strains opportunistically amplified under metabolic stress. These observations underscore the need to avoid oversimplified classifications of Lactobacillus as uniformly “beneficial” or “harmful,” instead advocating for functional subpopulation characterization through future strain-level analyses.

The Spearman correlation network we constructed revealed the multi-layer associations between differential bacterial genera and clinical metabolic indicators. In the Nor group, butyrate-producing bacteria including Faecalibacterium, Coprococcus, Lachnospiraceae ND3007 group, and Lachnospiraceae NK4A136 group were significantly enriched. These genera exhibited negative correlations with BMI, TG, and LDL-C, while demonstrating positive associations with HDL-C. These observations align with established evidence that elevated butyrate levels correlate with reduced adiposity and inverse associations with obesity markers and blood glucose levels (Amiri et al., 2023; Zhang et al., 2024). These findings are consistent with our results.

However, the Ruminococcus gnavus group in the Ow cohort exhibited inverse correlations with obesity-related metabolic markers (BMI, TC, and LDL-C) compared to the Nor group. This bacterial group has been widely recognized as a potential biomarker for multiple pathologies, including obesity, depression, and inflammatory bowel disease (Henke et al., 2019; Juge, 2023; He et al., 2025). Its elevated abundance in obese individuals aligns with existing evidence linking R. gnavus to adverse metabolic phenotypes, such as increased adiposity, elevated waist circumference, hypertriglyceridemia, and reduced HDL-C levels (Grahnemo et al., 2022). Notably, Hungatella exhibits a negative correlation with WBC counts, implying its potential protective role in modulating host inflammatory responses. This observation aligns with existing evidence demonstrating Hungatella’s capacity to enhance intestinal barrier function and immune regulation through SCFA production, which may suppress low-grade inflammation (Xing et al., 2025). While the exact mechanisms are not yet fully understood, these findings indicate that Hungatella may engage in compensatory upregulation of anti-inflammatory pathways in response to overweight conditions. This hypothesis merits additional research. Notably, the Escherichia−Shigella genus was significantly enriched in the Ob group and exhibited a positive correlation with BMI-a finding consistent with prior reports (Hong et al., 2024; Gao et al., 2025). While most studies associate Lactobacillus with metabolic benefits, our data revealed its positive correlation with Glu-AC and TG levels, aligning with limited but corroborative evidence (Li et al., 2022; Crudele et al., 2023; Feng et al., 2023). This discrepancy suggests strain- or host-specific metabolic effects of Lactobacillus. Conversely, Parabacteroides showed exclusive positive correlation with Glu-AC, potentially reflecting its dietary adaptation to high-fat/high-sugar conditions (de Wouters d'Oplinter et al., 2021).

Utilizing machine learning models, we successfully identified bacterial genera capable of distinguishing BMI stratification, with model AUC values ranging from 0.721 to 0.787. Notably, integrating these microbial markers with conventional clinical metabolic indicators (e.g., TG, HDL-C, UA) substantially enhanced diagnostic performance. It is important to highlight that, given that BMI has been utilized as the primary grouping variable in this research, this approach serves to mitigate the potential for pseudo-correlations arising from the interaction of cross-variables within the model, on the one hand, BMI alone often inadequately classifies obesity-related diseases at the individual level, as individuals with similar BMI values frequently exhibit divergent metabolic phenotypes (Coral et al., 2025). Therefore, we deliberately excluded BMI from the diagnostic model during the modeling process to test whether other metabolic indicators have independent stratification capabilities. The results demonstrated that key metabolic indicators (including TG, HDL-C, and UA) significantly enhanced the model’s discriminative performance independent of BMI, confirming their intrinsic predictive value for obesity risk stratification. Notably, when integrated with microbial markers, the combined model exhibited a substantial increase in AUC values, indicating that this “microbiome-metabolic” framework maintains robust diagnostic potential without BMI data. These findings highlight the clinical translatability of this integrated approach for non-invasive metabolic risk assessment.

While this study provides novel insights into structural variations of gut microbiota across different BMI categories, several limitations should be acknowledged. First, the cross-sectional design precludes causal inference between microbial alterations and metabolic indicators. While we propose that the microbial signatures identified in the overweight group hold potential for early risk stratification, this cross-sectional snapshot cannot demonstrate that these signatures actually predict future progression to obesity. Future studies incorporating longitudinal follow-up of overweight individuals are essential to validate the predictive power of these microbial markers for true early risk stratification. Such studies would track whether individuals with the ‘dysbiotic’ overweight profile we identified are indeed more likely to develop obesity and associated metabolic disorders over time. Second, the use of 16S rRNA sequencing limits taxonomic resolution and does not allow species- or strain-level differentiation. Thus, the observed associations—such as those involving Lactobacillus—may reflect underlying strain variability that cannot be resolved with 16S data alone. Future work using metagenomic approaches will be required to validate these findings and characterize functional differences more accurately. Third, key lifestyle factors (e.g., diet, physical activity, and sleep patterns) were not systematically incorporated into the analysis. These variables not only influence gut microbiota composition but may also confound the observed associations between BMI and microbial profiles. Future studies should employ comprehensive dietary frequency surveys and energy intake assessments, coupled with standardized exercise and sleep monitoring, to mitigate confounding effects and strengthen the model’s explanatory power. Additionally, the study’s single-center design and geographically restricted sampling may limit the generalizability of the findings. Given that gut microbiota composition is influenced by multifactorial determinants (e.g., genetics, diet, and environmental exposures), future multi-center studies involving diverse ethnic populations are warranted to validate and enhance the external validity of these conclusions.

This study revealed BMI-related alterations in gut microbiota composition and diversity, and demonstrated that integrating microbial features with clinical indicators enhances obesity stratification beyond BMI alone. These findings provide valuable reference evidence for the early stratification of obesity risk, suggesting that microbiota-informed assessment may complement traditional anthropometric measures. To advance toward practical application, future studies should include longitudinal cohorts, expanded sample sizes, and higher-resolution microbial profiling to clarify causal relationships and to determine whether these microbial signatures can reliably support early risk identification or guide individualized intervention strategies.