Polycystic ovary syndrome (PCOS) is a prevalent endocrine and metabolic disorder affecting approximately 11%–13% of women of reproductive age (1), characterized primarily by hyperandrogenism (HA), insulin resistance (IR), and ovulatory dysfunction, often accompanied by an increase in metabolic risks (1, 2). The diagnosis of PCOS remains fundamentally contentious, with the Rotterdam criteria serving as the primary diagnostic framework. These criteria require the presence of at least two of the following conditions: hyperandrogenism, ovulatory dysfunction, or polycystic ovarian morphology (PCOM). In contrast, the National Institutes of Health criteria are more stringent, emphasizing hyperandrogenism and ovulatory dysfunction while not mandating PCOM. Differences in the diagnostic criteria for PCOS not only reflect the complexity of its definition but also directly indicate the significant clinical heterogeneity of the disease: the Rotterdam criteria include more mild cases, whereas the NIH criteria focus on severe phenotypes. One of the important clinical manifestations of this heterogeneity lies in the differentiation of patients’ metabolic characteristics, particularly the significant differences in body mass index (BMI) and obesity status (2).

Traditionally, research and clinical management of PCOS have often used BMI as the primary reference indicator for phenotypic stratification. Despite its known limitations in assessing visceral fat distribution and metabolic health status, BMI, as a simple measure of body fat mass, cannot differentiate between fat and lean mass nor reflect fat distribution, particularly exhibiting inadequacies in evaluating central obesity. In analogous studies, phenotypes such as waist circumference, body fat percentage, and visceral fat area have been demonstrated to more accurately reflect metabolic risk and are closely associated with reproductive endocrine function (3). However, given that BMI is currently an internationally recognized assessment index and the vast majority of existing epidemiological and clinical mechanistic studies employ this metric for grouping, it remains a relatively feasible perspective for integrating evidence from the literature and conducting risk stratification at the present stage (4).

According to the World Health Organization criteria (overweight: BMI ≥ 25 kg/m²; obesity: BMI ≥ 30 kg/m²) (5). The prevalence of obesity among patients with PCOS ranges as high as 38%–88% (6), which is substantially higher than the 18% observed in general women of childbearing age (7). Epidemiological studies further confirm that for every 1% increase in obesity prevalence, the prevalence of PCOS rises by 0.4% (8).

Stratification studies based on BMI have further revealed that patients with PCOS exhibit significant differences in clinical phenotypes and metabolic risks: the obese phenotype of PCOS presents more severe metabolic abnormalities than the non-obese type, often accompanied by hypertension, type 2 diabetes mellitus, and an increased risk of cardiovascular diseases (6, 9), whereas the non-obese type is primarily characterized by reproductive endocrine dysfunction (9, 10). Although some normal-weight patients also exhibit metabolic abnormalities, obesity markedly exacerbates their pathological features. These clinical differences suggest that obese and non-obese PCOS may be driven by distinct pathophysiological mechanisms; however, the molecular basis underlying these related mechanisms remains incompletely elucidated. These unresolved mechanistic issues have sparked interest in the gut microbiota as a potential factor contributing to phenotypic variation among patients with polycystic ovary syndrome.

In recent years, the GM, as a key regulator of metabolism, inflammation, and hormonal homeostasis, has been regarded as an important link for explaining phenotypic heterogeneity in PCOS, particularly the differences between obese and non-obese phenotypes. The GM is acclaimed as the “second genome” of the human body; studies have shown that patients with PCOS exhibit evident dysbiosis, such as reduced α-diversity, elevated Firmicutes/Bacteroidetes ratio, and so forth (11). Further research reveals distinctions in microbial diversity, functional composition, and metabolic byproducts between obese and non-obese PCOS phenotypes. These differences participate in shaping the clinical phenotypes of the disease through multiple mechanistic layers, including inflammatory activation pathways, imbalanced short-chain fatty acid (SCFA) metabolism, bile acid metabolic disturbances, and aberrant signaling within the “gut–brain–ovary axis” (12–14). Therefore, analyzing GM characteristics from the perspective of BMI-based stratification not only facilitates elucidation of the complex pathological mechanisms underlying PCOS, but also provides novel insights for developing microbiota-targeted precision intervention strategies.

PCOS is a highly heterogeneous endocrine and metabolic disorder, with its clinical variations partially attributable to differences in BMI. As illustrated in Figure 1, obese PCOS patients typically present with pronounced IR, HA, and chronic inflammation, accompanied by elevated risks of metabolic syndrome and cardiovascular diseases. In contrast, non-obese patients, despite exhibiting less prominent metabolic abnormalities, may still experience ovulatory dysfunction and hormonal dysregulation (15), suggesting that these two phenotypes may stem from distinct biological underpinnings.

Epidemiological studies further quantified this difference. Research indicates that the prevalence of IR in obese PCOS patients can reach 70%–80% (16), significantly higher than the 15%–40% observed in non-obese counterparts (17, 18). Regarding HA, obese patients demonstrate more severe clinical manifestations (e.g., hirsutism, acne) and biochemical abnormalities (e.g., elevated free androgen index) (19). Furthermore, in terms of reproductive outcomes, obese PCOS patients face a markedly increased risk of infertility. A 2025 German cross-sectional study revealed that the infertility rate among obese PCOS patients was as high as 77.0%, compared to 53.0% in non-obese patients (20). Multiple large-scale studies have shown that the success rate of in vitro fertilization (IVF) in the high-BMI group is reduced by 20% to 40%, and live birth rate, number of oocytes retrieved, and embryo quality are all affected (21, 22). It is noteworthy that pre-pregnancy obesity itself is an independent risk factor for all pregnant women (including those without PCOS), significantly increasing the risks of GDM, gestational hypertension, preeclampsia, etc., and reducing IVF success rates, producing a synergistic effect when coexisting with PCOS. Another meta-analysis further confirmed that PCOS patients had a higher overall BMI (mean difference 1.76 kg/m²); even after adjusting for BMI, PCOS still independently increased the risk of GDM (OR 2.41) and preeclampsia (OR 2.30), suggesting that the pathophysiological abnormalities inherent to PCOS also contribute to the occurrence of pregnancy complications (23).

Further analysis revealed that this risk is dependent on BMI phenotype: in obese PCOS pregnant women, severe insulin resistance and metabolic disturbances are the main drivers of complications such as GDM and gestational hypertension, with high BMI amplifying insulin resistance and significantly exacerbating these risks (21). In non-obese individuals, although the degree of insulin resistance is relatively mild, hyperandrogenism itself is an independent risk factor for GDM and preeclampsia, potentially acting through mechanisms involving abnormal placental function (24). Table 1 shows the differences between obese and non-obese PCOS in terms of metabolism, endocrinology, and reproduction.

These differences in clinical and reproductive outcomes suggest that obese and non-obese PCOS may have distinct genetic bases. Genome-wide association studies (GWAS) have shown that both share some genetic variants and signaling pathways, but there are also phenotype-specific ones (25). Non-obese PCOS is more likely to exhibit specific associated loci, such as DENND1A (rs12000707), XBP1, and LINC02905 (which may affect follicular development via endoplasmic reticulum stress). In contrast, in the combined analysis of the obese type, insulin resistance–related variants (e.g., ERBB4) and another independent signal of DENND1A (rs569675099) are more prominent. These loci primarily influence insulin signaling, gonadotropin receptors, and follicular development pathways, highlighting the role of genetic heterogeneity in phenotypic differentiation.

Genetic factors do not act in isolation. Environmental and lifestyle factors such as diet, physical activity, and sleep can amplify genetic effects through gene–environment interactions (26). Patients with different BMI phenotypes show inconsistent treatment responses: the non-obese type is more sensitive to drugs (such as metformin) in terms of reproductive and hormonal improvements, with a greater reduction in testosterone levels compared to the obese type (27, 28). The obese type demonstrates more significant metabolic improvements (body weight, waist circumference, triglycerides, insulin sensitivity, etc.) following weight loss interventions (29, 30). This complementary advantage provides a basis for individualized management.

In summary, BMI classification reveals significant heterogeneity in PCOS at both clinical and genetic levels, and also provides a foundation for risk assessment and individualized management. However, genetic and metabolic differences cannot fully explain the clinical differentiation between the two phenotypes. In recent years, large-scale GWAS and Mendelian randomization (MR) research have shown that current evidence does not support a direct genetic association between specific PCOS susceptibility polymorphisms and gut microbiota typing (31, 32). Most MR effects were attenuated or disappeared after adjusting for BMI, insulin, and sex hormones (33), suggesting that the gut microbiota is more likely to indirectly influence phenotypes through metabolic and endocrine pathways, rather than being directly determined by a single susceptibility locus.In contrast, 16S rRNA and metagenomic sequencing studies have demonstrated that gut microbiota typing is more stably associated with phenotypes such as BMI, IR, and HA (34, 35), and is primarily driven by acquired factors including dietary patterns, obesity, and IR (35, 36). Based on the current data, a framework that better aligns with existing evidence is: the polygenic susceptibility of PCOS influences obesity, IR, and HA under the effect of adverse environmental exposures, subsequently shaping a specific microbial profile; the dysregulated gut microecology then feedbacks to exacerbate metabolic and hormonal abnormalities via pathways such as short-chain fatty acids, lipopolysaccharides, and bile acids. Genetic polymorphisms are more akin to relatively stable “underlying susceptibility,” while the gut microecology serves as a highly plastic “phenotypic amplifier.” Therefore, on the basis of clarifying the clinical and genetic heterogeneity related to BMI, further exploration of differences in microbial composition and function between obese and non-obese PCOS patients from the perspective of gut microecology will help comprehensively understand phenotypic differentiation and provide new targets for individualized intervention.

Existing evidence indicates that GM imbalance is a core pathological feature of PCOS(as shown in Figure 2), widely present in patients with different BMI phenotypes. However, obesity, as a significant metabolic state, not only further reshapes the microbial community structure but, more critically, acts as an “amplifier” of pathological pathways. The same microbial signals, in obese versus non-obese host environments, shape heterogeneous clinical phenotypes characterized by metabolic disorders or reproductive axis dysfunction through activation of downstream pathways with different weights.

With the increasingly clear role of the GM in the pathogenesis of PCOS, microecological modulation–based interventions have gradually emerged as a new area of research focus. Given the significant differences between obese and non-obese PCOS patients in terms of GM structure, metabolic burden, inflammatory status, and endocrine dysregulation, developing differentiated microecological intervention strategies guided by BMI stratification holds promise for achieving more precise and individualized disease management.

Compared with obese PCOS, research and evidence regarding the mechanisms of GM involvement in non-obese PCOS remain notably insufficient (2). Existing literature predominantly focuses on the GM dysbiosis characteristics of obese patients (42), whereas studies on non-obese PCOS are mostly at the exploratory stage, with limited sample sizes and inadequate stratification, resulting in a lack of systematic elucidation of its potentially specific mechanisms. For example, possible pathways in non-obese PCOS involving gut–brain axis signaling abnormalities, changes in specific microbial taxa, and GM-mediated modulation of HPO axis function via tryptophan metabolism currently yield inconsistent results due to study designs that are mainly cross-sectional and based on small samples, making it difficult to draw stable conclusions.

From a research perspective, the insufficiency of evidence is mainly reflected in the following aspects: the obese phenotype dominates research cohorts, and there is a lack of GM data stratified according to BMI or clinical subtypes (92); phenotypic characterization often relies on a binary definition of PCOS presence or absence, without adequately distinguishing reproductive-type from metabolic-type features (93); and GM is rarely incorporated into a unified framework integrating genetic and metabolic indicators (36). These factors collectively limit the ability to determine the position and causality of GM effects in non-obese PCOS, and to some extent marginalize management strategies for this population. In addition, current microecological intervention studies targeting different BMI phenotypes still lack specificity assessments and long-term efficacy data, further delaying the establishment of individualized treatment plans for non-obese PCOS.

Therefore, future research should treat non-obese PCOS as an independent subject of investigation, incorporate BMI and clinical subtype stratification in cohort construction, and combine longitudinal follow-up or interventional designs to systematically evaluate the dynamic relationship between GM and hormonal and metabolic indicators. Simultaneously, GM should be integrated into multi-omics frameworks and validated across diverse populations and dietary backgrounds, so as to progressively fill the evidence gap regarding GM in non-obese PCOS and develop more targeted management and intervention strategies for this subgroup (2).

The clinical heterogeneity of PCOS has been widely recognized; however, existing classification systems primarily rely on horizontal categorization based on clinical manifestations, such as hyperandrogenemia, ovulatory dysfunction, or PCOM, while systematic integration of metabolic background differences among patients remains relatively insufficient. This review, based on a comprehensive analysis stratified by BMI, demonstrates that BMI not only reflects variations in body fat distribution and metabolic status but also participates, to a certain extent, in shaping the pathophysiological patterns dominated by different PCOS phenotypes. Integrating the available evidence, the core argument proposed in this review is that BMI is not merely a clinical stratification marker for PCOS but serves as a critical determinant that drives phenotype-specific alterations in gut microbiota by shaping distinct adverse biological pathways—either metabolism–inflammation–dominant or reproductive axis dysfunction–dominant—thereby providing a biological rationale for stratified intervention strategies.

Based on current evidence, obese PCOS is typically characterized by insulin resistance, hyperandrogenism, and metabolic syndrome, with more pronounced gut microbiota dysbiosis manifested as reduced microbial diversity, enrichment of pro-inflammatory taxa, and disturbances in short-chain fatty acid and bile acid metabolism. These alterations further exacerbate metabolic burden through pathways involving chronic low-grade inflammation, abnormal energy metabolism, and endocrine regulatory imbalance. In contrast, non-obese PCOS patients exhibit relatively milder overall metabolic disturbances, yet specific changes in gut microbiota composition and function persist; these changes may influence hormonal regulation and ovulatory function via the gut–brain–ovary axis. Such differences suggest that alterations in gut microbiota across BMI phenotypes are not simply quantitative distinctions but may contribute to pathophysiological divergence determining the dominant disease pathways. It should be noted that metabolic, inflammatory, and neuroendocrine pathways are not mutually exclusive but exhibit varying relative weights within different BMI-associated metabolic contexts.

On this basis, this review systematically integrates the association between PCOS and gut microbiota from the perspective of BMI stratification. Unlike previous reviews that primarily analyzed the overall PCOS population and focused mainly on global microbiota dysbiosis, this review expands current understanding in the following aspects: (1) systematic integration of BMI stratification, clinical phenotypic characteristics, and differences in gut microbiota composition and function; (2) recognition of non-obese PCOS as a subtype with independent biological features, rather than a milder manifestation of obesity-related phenotypes; (3) emphasis on the pivotal role of BMI-associated metabolic environments in shaping microbiota characteristics and their downstream metabolic–endocrine effects, linking these to stratified intervention strategies. This perspective facilitates a deeper understanding of the biological basis underlying PCOS phenotypic heterogeneity.

Despite the growing number of studies on the relationship between PCOS and gut microbiota in recent years, several limitations warrant attention. First, most studies have limited sample sizes and are predominantly cross-sectional in design, making it difficult to establish causal relationships between gut microbiota alterations and disease phenotypes. Second, research subjects are mainly concentrated in Asian and European–American populations; differences in the metabolic significance of BMI across ethnicities and dietary backgrounds may affect the generalizability of findings. Additionally, existing studies tend to focus on obese PCOS, while systematic evidence regarding microbiota-specific mechanisms and clinical implications in non-obese PCOS remains lacking (see Section VI for details).

Future research should conduct more refined longitudinal and interventional studies based on BMI stratification to dynamically assess the relationship between gut microbiota changes and metabolic and reproductive outcomes. Concurrently, integration of multi-omics data will help clarify the interactions among BMI, gut microbiota, and PCOS phenotypes. Building on these insights, formulating differentiated microecological intervention strategies according to BMI phenotypes may enhance the precision and clinical feasibility of individualized management for PCOS.