Correlation between intestinal flora and glucuronidase indexes and sex hormone levels and their diagnostic efficacy evaluation for premature ovarian failure.

This study was approved by the Ethics Committee of Jiangxi Maternal and Child Health Hospital (Approval No. JXMCHH-2021-POF01). All procedures performed in studies involving human participants were conducted in accordance with the ethical standards of the institutional and national research committee and with the Helsinki Declaration and its later amendments or comparable ethical standards. Written informed consent was obtained from all individual participants included in the study. The sample size was determined using the formula for estimating a proportion with a specified margin of error: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\text{sample\:size}=\frac{{Z}^{2}\times\:P\times\:(1-P)}{{E}^{2}}$$\end{document} Here, P represents the incidence rate of POF, E the acceptable margin of error, and Z the confidence level. Based on a reported incidence of 3.5% [ 12 ], we applied P = 0.035, E = 0.05, and Z = 1.96 (corresponding to a 95% confidence level). The calculated sample size was 52 cases. To account for an anticipated 20% dropout rate, the final sample size was increased to 62 cases. Inclusion criteria: Participants in the POF group were selected based on widely accepted diagnostic standards for premature ovarian failure (also referred to as primary ovarian insufficiency), as recommended by the European Society of Human Reproduction and Embryology (ESHRE) and The American College of Obstetricians and Gynecologists (ACOG). Specifically, patients met the following criteria: (1) Age ≤ 40 years, consistent with the clinical definition of POI. (2) A history of previously regular menstruation followed by amenorrhea lasting ≥ 4 months prior to consultation, in accordance with ESHRE and ACOG guidelines, and an FSH level ≥ 40 IU/L) [ 13 ]. (3) Demonstrated good compliance and cooperation with all required examinations. (4) Presence of an intact uterus with an endometrial thickness ≤ 5 mm. (5) POF not attributed to factors such as genetic history or immune diseases. Exclusion criteria: Patients were excluded if they had any of the following: (1) History of radiotherapy or chemotherapy. (2) Receipt of hormone therapy within the past 90 days. (3) Coexisting endocrine disorder such as diabetes mellitus, hyperthyroidism or hypothyroidism. (4) Use of medications affecting the intestinal microbiota (e.g., antibiotics or probiotics) within the past 90 days. (5) Ovarian disorders such as polycystic ovary syndrome and ovarian cysts. (6) Patients with autoimmune diseases. The control group consisted of 52 healthy women who were evaluated over the same time, while the 62 patients with POF who satisfied the inclusion and exclusion criteria between April 2022 and April 2024 were chosen as the POF group. Women who were pregnant or nursing were excluded because pregnancy and lactation are associated with significant physiological changes in hormonal profiles, which could confound the interpretation of sex hormone levels (such as AMH, FSH, LH, and E2) and their relationship with intestinal flora and glucuronidase activity. This exclusion ensures that the control group represents the baseline hormonal status of healthy, non-pregnant women. General demographic and clinical data collected included age, body mass index (BMI), age at menarche, number of pregnancies, and history of mechanical abortion. Serum samples were collected on the 3rd to 5th days of the menstrual cycle for the control group, corresponding to the early follicular phase when gonadotropin and estrogen levels are relatively stable. For women with POF, serum samples were collected at the time of clinical evaluation, as the majority presented with amenorrhea and did not have a predictable cycle. This difference in sampling timing was unavoidable due to the pathophysiology of POF but represents a potential source of variability in hormone measurements. To account for this, we performed sensitivity analyses stratifying patients according to amenorrhea duration (< 6 months vs. ≥6 months), and results are provided in Supplementary Table S4. The blood was spun at 5000 rpm for 10 min with a centrifugation radius of 11.5 cm after standing at room temperature for 30 min. AMH was found using an enzyme-linked immunosorbent test after the supernatant was obtained. Chemiluminescence was used to quantify LH, FSH, and E2. Stool: Fresh stool samples from women in the POF group and the CG were collected outside the menstrual cycle, stored in sterile centrifuge tubes, and preserved at −80 °C. 0.5 g of stool sample was added to 0.1 mol/L PBS and mixed thoroughly. The mixture was centrifuged at an ultra-low temperature, and the supernatant was collected. Following three iterations of this procedure, the final supernatant was moved to an EP tube and centrifuged for ten minutes at an ultra-low temperature at 12,000 r/min. The resulting precipitate was collected for deoxyribonucleic acid (DNA) extraction using a fecal DNA purification kit, following the manufacturer’s instructions. A nucleic acid protein detector was used to test the concentration of the purified DNA. Using 0.8% agarose gel electrophoresis, the size and integrity of the isolated DNA fragments were evaluated, and a gel imaging equipment was used to visualize the results. Quantitative real-time PCR (qPCR) was performed to determine bacterial copy numbers for Lactobacillus , Bifidobacterium , Enterococcus , and Escherichia coli . Primer and probe sequences used for the amplification of 16 S rRNA genes are detailed in Supplementary Table S1 . Each reaction used 10 µL of SYBR Green Master Mix, 0.5 µL each of forward and reverse primers (10 µM), 1 µL DNA template, and nuclease-free water to a final volume of 20 µL. The qPCR protocol was as follows: initial denaturation at 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s, 60 °C for 10 s, and 68 °C for 30 s. Standard curves were generated using 10-fold serial dilutions of plasmids containing the target bacterial 16 S rRNA gene fragments, with copy numbers ranging from 10⁶ to 10¹ CFU/µL. Ct values were converted to log CFU/g based on these standard curves. The slope of the standard curves ranged from − 3.32 to −3.47, with R² values exceeding 0.995, indicating high amplification efficiency. The limit of detection (LOD) was approximately 100 copies/reaction and the limit of quantification (LOQ) was 500 copies/reaction. To enhance the representativeness and universality of IF in diagnosing POF, healthy women were selected as controls. The study focused on the most common and representative intestinal microorganisms, including Escherichia coli , Lactobacillus , Enterococcus , and Bifidobacterium , for comparative analysis. gmGUS Activity: The activity of gmGUS in stool samples was measured using a colorimetric enzyme activity assay kit (Sigma-Aldrich, Catalog No. MAK174), based on p-nitrophenyl-β-D-glucuronide as the substrate. The reaction was quantified by measuring absorbance at 405 nm. The intra-assay coefficient of variation (CV) was 4.3%, and the inter-assay CV was 6.7%. The normal reference range (0–2486 U/mL) was derived from the manufacturer’s instructions. (Fig. 1 ) Fig. 1 Study flow chart Study flow chart All statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). The Kolmogorov–Smirnov test was used to assess the normality of continuous variables. Variables following a normal distribution were expressed as mean ± standard deviation (SD) and compared using independent-sample t-tests. Non-normally distributed variables were presented as median (interquartile range, IQR) and analyzed using the Mann–Whitney U test. Categorical variables were expressed as counts and percentages. The chi-squared (χ²) test was used for categorical comparisons when all expected cell counts were ≥ 5. In cases where more than 20% of expected cell counts were < 5, or any expected frequency was < 1, the Fisher’s exact test was used instead, in accordance with standard statistical guidelines. Normality of continuous variables was assessed using the Kolmogorov–Smirnov (K–S) test. Test statistics and p-values for each variable are presented in Supplementary Table S2. Variables with non-normal distributions ( P  < 0.05) were further analyzed using Spearman’s rank correlation coefficient to assess associations. Both Pearson and Spearman correlation coefficients are reported where appropriate. Receiver Operating Characteristic (ROC) curve analysis was used to evaluate the diagnostic performance of individual and combined biomarkers. The area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. Logistic regression models were constructed to assess the association between gut microbial variables and POF risk. Model outputs included regression coefficients, standard errors, odds ratios (OR), 95% confidence intervals (CI), and p-values. Multicollinearity was assessed using variance inflation factors (VIF), with VIF > 5 considered indicative of collinearity. Model calibration was examined using the Hosmer–Lemeshow test, the Brier score, calibration slope, and intercept. A calibration plot was generated to visualize agreement between predicted and observed probabilities (Supplementary Figure S1 ). A two-sided p-value of < 0.05 was considered statistically significant.

A total of 62 women with POF and 52 age-matched healthy controls were included in the study. Baseline clinical parameters including age, BMI, age at menarche, number of pregnancies, history of induced abortion, history of gastric ulcer, and history of Helicobacter pylori infection showed no statistically significant differences between the POF group and controls (all P  > 0.05; Table 1 ). This indicates that the two groups were comparable in terms of demographic and reproductive history variables. Table 1 Basic characteristics of subjects in the two groups Characteristics POF group ( n  = 62) Control group( n  = 52) t/Z P Age (years) 32.83 ± 5.31 31.28 ± 5.27 1.558 0.122 BMI(kg/m²) 22.38 ± 2.42 22.40 ± 2.37 0.044 0.965 Age at menarche (years) 13.27 ± 2.38 13.11 ± 2.35 0.360 0.720 Pregnancy history 1.391 0.164  1 time 47(75.81) 45(86.54)  2 times 10(16.13) 4(7.69) more than 2 times 5(8.06) 3(5.77) History of induced abortion 1.577 0.115  None 38(61.29) 40(76.92)  1 time 20(32.26) 8(15.38)  2 times 3(4.84) 2(3.85)  > 2 times 1(1.62) 2(3.85) History of gastric ulcer 2(3.22) 1(1.92) 0.187 0.665 History of Helicobacter pylori infection 1(1.61) 0(0.00) 0.846 0.358 History of gastrointestinal tumors 3(4.84) 0(0.00) 2.584 0.108 Basic characteristics of subjects in the two groups Quantitative analysis of fecal samples demonstrated significant alterations in gut microbiota among women with POF. Compared with controls, the POF group had markedly lower abundances of Lactobacillus (8.27 ± 0.95 vs. 10.24 ± 1.24 log CFU/g, P  < 0.001), Bifidobacterium (8.18 ± 0.89 vs. 10.17 ± 1.21 log CFU/g, P  < 0.001), and gmGUS activity (481.82 ± 34.91 vs. 1742.17 ± 141.65 U/mL, P   0.05, Table 2 ). Table 2 Intestinal flora and GmGUS activity in the POF and the control groups Indicator POF Group ( n  = 62) Control Group ( n  = 52) t-value P -value Lactobacillus (log CFU/g) 8.27 ± 0.95 10.24 ± 1.24 9.597 < 0.001 Enterococci (log CFU/g) 9.15 ± 1.15 8.79 ± 1.02 1.752 0.083 Bifidobacterium (log CFU/g) 8.18 ± 0.89 10.17 ± 1.21 10.099 < 0.001 Escherichia coli (log CFU/g) 10.26 ± 1.58 9.79 ± 1.27 1.727 0.087 gmGUS (U/mL) 481.82 ± 34.91 1742.17 ± 141.65 46.732 < 0.001 Intestinal flora and GmGUS activity in the POF and the control groups Women with POF exhibited profound alterations in sex hormone profiles compared with healthy controls. Specifically, serum AMH and E2 levels were significantly reduced (AMH: 2.38 ± 0.44 vs. 12.73 ± 2.91 ng/mL; E2: 45.28 ± 6.27 vs. 179.13 ± 28.64 pmol/L; both P  < 0.001). Conversely, FSH and LH levels were markedly elevated (FSH: 73.72 ± 10.28 vs. 6.28 ± 1.23 U/mL; LH: 42.84 ± 5.38 vs. 4.16 ± 0.85 U/L; both P  < 0.001; Table 3 ). These hormonal changes confirm the impaired ovarian function in the POF cohort. Table 3 Hormone levels in the POF and control groups Indicator POF Group ( n  = 62) Control Group ( n  = 52) t-value P -value AMH (ng/mL) 2.38 ± 0.44 12.73 ± 2.91 27.654 < 0.001 FSH (U/mL) 73.72 ± 10.28 6.28 ± 1.23 46.993 < 0.001 LH (U/L) 42.84 ± 5.38 4.16 ± 0.85 51.275 < 0.001 E2 (pmol/L) 45.28 ± 6.27 179.13 ± 28.64 35.819 < 0.001 Hormone levels in the POF and control groups Pearson correlation analysis within the POF group revealed significant associations between gut microbiota, gmGUS activity, and sex hormones (Fig. 2 ). Lactobacillus showed a negative correlation with AMH ( r = −0.323, P  < 0.05) and E2 ( r = −0.402, P  < 0.001), positively correlation with FSH ( r  = 0.365, P  < 0.05) and LH ( r  = 0.411, P  < 0.001). Bifidobacterium exhibited a negative correlation with AMH ( r = −0.321, P  < 0.05) and E2 ( r = −0.362, P  < 0.05), and a positive correlation with FSH ( r  = 0.431, P  < 0.001) and LH ( r  = 0.353, P  < 0.05). gmGUS activity was negatively correlated with AMH ( r = −0.402, P  < 0.001), FSH ( r = −0.423, P  < 0.001), and LH ( r = −0.392, P  < 0.001); positively correlated with E2 ( r  = 0.374, P  < 0.05). These findings suggest a potential mechanistic link between intestinal dysbiosis, altered gmGUS activity, and hormonal imbalance in POF. Fig. 2 Heatmap of Pearson’s correlation coefficients between intestinal flora and sex hormone levels in POF patients( n  = 62). Rows represent intestinal flora components, and columns represent sex hormones. The color bar, labeled “Correlation Coefficient (Pearson’s r),” ranges from − 0.4 to 0.4: green indicates negative correlations, red indicates positive correlations, and white represents no correlation (near 0). Pearson’s correlation coefficients range from − 1 to 1, where − 1 indicates a perfect negative correlation, 0 indicates no correlation, and 1 indicates a perfect positive correlation. Units: intestinal flora expressed as log CFU/g, gmGUS as U/mL, and hormones as ng/mL or pmol/L as appropriate. Statistical test: Pearson’s correlation (two-tailed); significance indicated as * P  < 0.05 and ** P  < 0.001 Heatmap of Pearson’s correlation coefficients between intestinal flora and sex hormone levels in POF patients( n  = 62). Rows represent intestinal flora components, and columns represent sex hormones. The color bar, labeled “Correlation Coefficient (Pearson’s r),” ranges from − 0.4 to 0.4: green indicates negative correlations, red indicates positive correlations, and white represents no correlation (near 0). Pearson’s correlation coefficients range from − 1 to 1, where − 1 indicates a perfect negative correlation, 0 indicates no correlation, and 1 indicates a perfect positive correlation. Units: intestinal flora expressed as log CFU/g, gmGUS as U/mL, and hormones as ng/mL or pmol/L as appropriate. Statistical test: Pearson’s correlation (two-tailed); significance indicated as * P  < 0.05 and ** P  < 0.001 The K–S test results indicated that several variables, including fecal gmGUS activity and the relative abundance of Enterococcus and E. coli, deviated from normal distribution ( p  < 0.05). Pearson and Spearman correlation analyses between microbiota-related variables and serum hormone levels are summarized in Supplementary Table S3. Despite minor differences, the direction and significance of correlations were largely consistent between methods, supporting the robustness of the findings. The full logistic regression results are presented in Supplementary Table S5. All predictors demonstrated VIF values < 2, confirming no significant collinearity. Calibration metrics showed good model performance (Hosmer–Lemeshow p  = 0.47; Brier score = 0.063; calibration slope = 0.94; intercept = 0.05). The calibration plot (Supplementary Figure S1 ) further confirmed close alignment between predicted and observed risk probabilities, supporting the robustness of the regression model. ROC curve analysis was performed to evaluate the diagnostic potential of individual microbial markers and gmGUS activity (Fig. 3 ; Table 4 ). Lactobacillus yielded an AUC of 0.770 (95%CI: 0.680–0.866), with sensitivity 80.6% and specificity 82.7%. Bifidobacterium achieved an AUC of 0.758(95%CI: 0.667–0.849), with sensitivity 71.0% and specificity 80.8%. Similarly, gmGUS activity had an AUC of 0.770(95%CI: 0.683–0.857), with 82.3% sensitivity and 69.2% specificity. Fig. 3 ROC curves evaluating the diagnostic efficacy of individual intestinal flora and gmGUS activity for POF. Sample size: POF group (n = 62), Control group (n = 52) ROC curves evaluating the diagnostic efficacy of individual intestinal flora and gmGUS activity for POF. Sample size: POF group (n = 62), Control group (n = 52) Table 4 Efficacy analysis of intestinal flora and GmGUS activity in diagnosing POF Intestinal flora Cutoff value(unit) AUC P 95%CI Youden Index Sensitivity (%) Specificity (%) Lactobacillus 9.024 log CFU/g 0.773 <0.001 0.680–0.866 0.663 0.806 0.827 Bifidobacteriu m 8.723 log CFU/g 0.758 <0.001 0.667–0.849 0.517 0.710 0.808 gmGUS 403.733U/mL 0.770 <0.001 0.683–0.857 0.472 0.823 0.692 Efficacy analysis of intestinal flora and GmGUS activity in diagnosing POF A logistic regression model incorporating Lactobacillus, Bifidobacterium, and gmGUS activity was constructed. The model demonstrated excellent diagnostic accuracy with AUC = 0.912 (95% CI: 0.862–0.962), sensitivity = 95.2%, and specificity = 71.2% (Fig. 4 ). To assess clinical applicability, PPV and NPV were calculated at different assumed population prevalences of POF. At a prevalence of 1%, the PPV was 3.2% and the NPV was 99.9%. At 5% prevalence, the PPV increased to 15.0% and NPV was 99.0%. At 10% prevalence, the PPV reached 26.4% and NPV was 97.5% (Supplementary Table S6). These findings indicate that while the model excels in ruling out disease (high NPV), its positive predictive utility depends strongly on underlying prevalence. Fig. 4 ROC curve for the combined logistic regression model incorporating Lactobacillus, Bifidobacterium, and gmGUS activity. Sample size: POF group (n = 62), Control group (n = 52) ROC curve for the combined logistic regression model incorporating Lactobacillus, Bifidobacterium, and gmGUS activity. Sample size: POF group (n = 62), Control group (n = 52)

This study provides novel evidence that women with POF exhibit significant alterations in intestinal flora, specifically reduced levels of Lactobacillus and Bifidobacterium , and decreased gut microbial gmGUS activity. These microbial markers are closely associated with abnormal sex hormone profiles, including decreased AMH and E2, and increased FSH and LH. Importantly, our combined diagnostic model incorporating gmGUS, Lactobacillus , and Bifidobacterium achieved high diagnostic accuracy (AUC = 0.912, sensitivity = 95.2%, specificity = 71.2%), with an exceptionally high negative predictive value (NPV = 99.9%), indicating strong potential as a non-invasive screening tool to rule out POF in clinical settings. However, these findings should be interpreted within the context of the study’s limitations, including its single-center design, modest sample size, and the absence of dietary, metabolomic, or in vivo validation data. In conclusion, this study supports the role of the gut–ovary axis in POF and proposes a promising diagnostic adjunct based on gut microbiota and enzymatic markers. Future large-scale, multicenter studies with mechanistic exploration are needed to validate these findings and explore potential microbiota-targeted interventions for POF.

POF is a clinical condition characterized by early cessation of ovarian function and estrogen deficiency, with increasing recognition of its association with systemic and environmental factors [ 14 , 15 ]. In this study, we observed significantly reduced levels of Lactobacillus , Bifidobacterium , and gmGUS activity in women with POF compared to healthy controls. Additionally, serum AMH and E2 levels were markedly decreased, while FSH and LH were elevated in the POF group. Correlation analyses revealed distinct relationships between intestinal flora, gmGUS activity, and hormone levels, and the diagnostic model combining microbial and enzymatic indicators demonstrated high accuracy (AUC = 0.912). These findings suggest that intestinal flora and its functional enzyme gmGUS may serve as promising non-invasive biomarkers for early identification of POF. In recent years, studies have found that the biological aging of women in their thirties and fifties is significantly accelerated, which coincides with the time point of declining fertility and menopause [ 16 ]. As life expectancy has increased globally, ovarian ageing has progressively emerged as a significant health issue for women. Ovarian aging is very complex and is not only related to changes in sex hormone levels [ 17 , 18 ]. Ovarian function is related to intestinal flora, and the intestinal flora-ovarian axis contributes to follicle development. There is an interaction between the composition and functional characteristics of the intestinal flora and ovarian function [ 19 ]. The outcome of this investigation found that the levels of Lactobacillus , Bifidobacterium , and gmGUS activity in the POF group were lower than those in the CG. There was no discernible variation in enterococci and Escherichia coli involving the two groupings, indicating that POF patients have already developed abnormal intestinal flora. Because lactobacillus and bifidobacterial are beneficial bacteria that have a significant part in maintaining intestinal health, enhancing immune function, and regulating metabolism. POF is often accompanied by chronic inflammation. An imbalance in Lactobacillus and Bifidobacterium populations contributes to the ovarian immune-inflammatory response, resulting in the increased synthesis of intestinal mucosal secretory immunoglobulins. In addition, these altered microbial communities interact with intestinal mucosal immune cells to exacerbate the disruption of the delicate equilibrium between the ovarian environment and the intestinal milieu [ 20 – 22 ]. At the same time, the interaction between intestinal flora and estrogen needs to be considered. Intestinal flora affects estrogen levels, and intestinal epithelial estrogen β receptors also affect the diversity of intestinal flora [ 23 , 24 ]. gmGUS activity mainly affects estrogen metabolism, which can mediate estrogen dissociation, accelerate its intestinal reabsorption, and release it into the circulatory system [ 25 ]. The active, dissociated form of the hormone binds to estrogen receptor α, regulating multiple intracellular signaling pathways. This interaction modulates the transcriptional activity of estrogen-responsive genes, initiating a cascade of molecular events that support ovarian health and maintain physiological homeostasis [ 26 , 27 ]. The findings of this study further demonstrated that AMH and E2 showed a negative correlation with Lactobacillus , however, FSH and LH showed a positively correlation. Similarly, Bifidobacterium exhibited negative correlations with AMH and E2 and positive correlations with FSH and LH. In contrast, gmGUS activity was negatively correlated with AMH, FSH, and LH, but positively correlated with E2. These results align with previous studies, which suggest that intestinal flora can exert immunomodulatory effects, thereby influencing the production and metabolism of estrogen [ 28 , 29 ]. Previous study has found that in rats with a menopausal model, estrogen levels decreased significantly after the intestinal flora of the rats was adjusted, and long-term supplementation of lactobacillus and β-glucan can prevent menopausal symptoms induced by estrogen deficiency [ 30 ]. This study preliminarily suggests that alterations in intestinal flora and gmGUS activity may influence hormone balance and represent one of the mechanisms contributing to the development of POF. The mean AMH level observed in the POF group (2.38 ± 0.44 ng/mL) was higher than values typically reported in advanced-stage POF, where AMH is often < 0.5 ng/mL or undetectable [ 31 ]. This discrepancy likely reflects the inclusion of patients in the early stages of ovarian insufficiency or those who retained partial ovarian function. As our inclusion criteria were based on clinical features (amenorrhea ≥ 4 months and FSH ≥ 40 IU/L) in line with ESHRE and ACOG recommendations, we captured a clinically relevant spectrum of POF rather than only end-stage cases. This phenotypic variability may explain the observed AMH levels and reinforces the real-world applicability of our diagnostic model [ 14 ]. Nevertheless, future studies should consider stratifying participants based on AMH thresholds to assess the performance of intestinal flora and gmGUS activity markers across different severities of ovarian insufficiency. To clarify whether the intestinal flora combined with gmGUS activity has a diagnostic effect on POF, we established ROC curve analysis separately and in combination. The sensitivity and specificity of the intestinal flora combined with gmGUS activity was increased to 0.952 and 0.712, respectively, implying that the model can diagnose POF in some way. The sensitivity and specificity of the intestinal flora combined with gmGUS activity in the diagnosis of POF are improved, probably because the two are complementary in reflecting the functional status of the ovaries and can capture disease-related characteristics more comprehensively. This result suggests that the intestinal flora and its functional index gmGUS may be an important diagnostic tool for POF and establishes a foundation for further research on the relationship between intestinal flora and ovarian function. Although the diagnostic model demonstrated high sensitivity (95.2%) and an excellent NPV (99.9%), its relatively low PPV (3.2%) reflects the low prevalence of POF in the general population. Accordingly, the model is best applied as a preliminary screening tool to rule out POF in women presenting with menstrual irregularities or suspected ovarian dysfunction. In higher-risk populations, such as individuals with a family history of POF, autoimmune disease, or prior ovarian surgery, the PPV is expected to be higher. Thus, this model shows promise as a non-invasive adjunct to conventional diagnostic approaches, particularly when combined with hormonal assays in clinical practice. Although the correlation analysis was performed solely within the POF group to assess relationships between intestinal flora, gmGUS, and sex hormone levels in the disease state, the broader diagnostic utility of these findings is supported by between-group comparisons and multivariate modeling. Specifically, microbial, and enzymatic markers that were significantly altered in POF compared to controls ( Lactobacillus , Bifidobacterium , and gmGUS activity) also demonstrated meaningful correlations with hormone dysregulation unique to POF. This dual-level evidence enhances the biological plausibility of these markers and justifies their inclusion in a diagnostic framework. The combined diagnostic model, developed from both POF and control data, showed robust performance (AUC = 0.912), supporting the translational value of these intra-group correlations for broader diagnostic application. The high diagnostic accuracy of the combined model suggests that Lactobacillus , Bifidobacterium , and gmGUS could serve as non-invasive biomarkers for POF, complementing current hormonal assays [ 32 ]. Early detection could enable timely interventions to mitigate risks like osteoporosis and cardiovascular disease [ 33 ]. Additionally, these findings open avenues for therapeutic strategies, such as probiotics or dietary interventions to modulate gut flora, as explored in Journal of Clinical Endocrinology & Metabolism for hormonal balance [ 34 ]. The study’s single-center design and small sample size may limit the generalizability of findings on the relationship between intestinal flora, gmGUS, and POF. Although patients with diabetes and other endocrine disorders were excluded, endometriosis, a condition associated with gut microbiota alterations that influence hormonal balance and immune function [ 35 ], was not assessed. In addition, environmental factors such as diet and alcohol consumption, both established modulators of the gut microbiota [ 36 ], were not evaluated. Data on dietary habits or gastrointestinal diseases beyond the exclusion criteria were not collected, which may have introduced residual confounding. In particular, dietary patterns, known to strongly influence gut microbial composition and gmGUS activity, were not accounted for in this study. Fiber-rich diets tend to promote the growth of Lactobacillus and Bifidobacterium , whereas high-fat or protein-rich diets may suppress these taxa and influence systemic inflammation. Moreover, specific dietary components such as phytoestrogens and polyphenols can alter estrogen metabolism through their effects on gmGUS activity. Thus, the lack of dietary control introduces potential confounding in the interpretation of gut microbiota and enzymatic differences between groups. Future research should include detailed dietary assessments to delineate the contribution of nutritional factors to the gut–ovary axis in POF. The absence of metabolomic or estrogen metabolite data restricts mechanistic confirmation of gmGUS’s role in estrogen bioavailability. Additionally, the interactions between intestinal flora, gmGUS, and POF lack validation through in vivo or in vitro studies. Future research must address these limitations to enhance and substantiate these findings. The findings of this study carry important clinical implications. The observed alterations in gut microbiota and gmGUS activity suggest that these markers may serve as non-invasive adjunctive tools for the early identification of POF, complementing standard hormonal assays. The high sensitivity and negative predictive value of the combined diagnostic model highlight its potential utility as a screening approach to rule out POF in women presenting with menstrual irregularities, thereby reducing reliance on invasive or repeated hormonal testing. Moreover, the associations between microbiota, gmGUS, and sex hormone levels raise the possibility of therapeutic strategies, including probiotics, prebiotics, or dietary interventions aimed at restoring microbial balance and supporting ovarian function. Importantly, the study’s limitations should be considered. These include the single-center design, relatively modest sample size, absence of dietary and metabolomic data, and lack of in vivo or mechanistic validation. These factors may limit the generalizability and causal interpretation of the findings. One methodological consideration of this study is the timing of hormone sampling. In controls, blood samples were collected during the early follicular phase (day 3–5), which is the standard timing for evaluating baseline gonadotropin and estradiol levels. However, most POF patients presented with prolonged amenorrhea, precluding standardized cycle-based sampling. As such, their samples were obtained at the time of evaluation, which may introduce variability in hormone concentrations. Although our sensitivity analysis stratified by duration of amenorrhea showed no material differences in the correlations between intestinal flora, gmGUS activity, and hormone levels (Supplementary Table S4), this limitation should be acknowledged. Future studies should attempt to harmonize sampling timing across groups or apply cycle-independent biomarkers to minimize this source of confounding. Although the combined model demonstrated high diagnostic accuracy and outstanding negative predictive value, the positive predictive value remained low under low-prevalence assumptions, reflecting the rarity of POF in the general population. This highlights an inherent limitation of predictive values calculated from case–control data, which do not reflect real-world prevalence. Accordingly, PPV and NPV reported here should be interpreted with caution and considered illustrative rather than definitive. These results emphasize the importance of validating the model in prospective, population-based cohort studies, where true prevalence and predictive values can be accurately estimated. External validation in larger, multicenter cohorts is essential before clinical implementation. Future research should focus on validating these findings in larger, multicenter cohorts with diverse demographics and disease severities, incorporating mechanistic studies to clarify the causal role of microbiota–hormone interactions. Interventional trials assessing microbiota-targeted therapies could further determine whether modulation of the gut–ovary axis may delay disease progression or improve reproductive and metabolic outcomes in women at risk of POF.

Premature ovarian failure (POF), also known as primary ovarian insufficiency (POI), refers to the loss of normal ovarian function before the age of 40. It affects approximately 1–5% of women worldwide and is characterized by amenorrhea, hypoestrogenism, and elevated gonadotropins, leading to infertility and long-term health consequences such as cardiovascular disease, osteoporosis, and psychological distress [ 1 – 3 ]. Despite its significant burden on reproductive health and quality of life, the underlying pathogenesis of POF remains poorly understood, and early diagnosis remains a challenge. The etiology of POF is complex and multifactorial. Primary POF has been linked to genetic factors, enzyme deficiencies, autoimmune diseases, radiation damage, and chemical exposures. Iatrogenic POF is commonly linked to surgical interventions, chemotherapy, and radiotherapy [ 4 , 5 ]. Understanding the underlying causes and mechanisms of POF is crucial for developing strategies for early warning, screening, and early diagnosis, and treatment in high-risk populations. Recent research highlights the significant role of intestinal flora (IF) and its microbial composition in female reproductive endocrine diseases, suggesting its potential involvement in the pathogenesis of POF [ 6 ]. The IF is a complex symbiotic community of microorganisms inhabiting the human gut, distinguished by its vast abundance, extensive genetic diversity, and functional versatility. It plays a pivotal role in regulating key physiological systems, including metabolism, immune responses, and endocrine function. Under healthy conditions, IF maintains a stable equilibrium, supporting a balanced and mutually beneficial microbiome–host relationship [ 7 , 8 ]. Graham ME et al. pointed out that the intestinal microbial changes in patients with premature ovarian dysfunction are mainly manifested in the decrease of Bacteroidetes, Bifidobacterium, etc [ 9 ]. Other studies have shown that alterations in internal factors, particularly sex hormone levels, and the activity of gut microbial β-glucuronidase (gmGUS), can convert estrogen from its inactive to active form, thereby influencing systemic estrogen concentrations and impairing follicular development. Conversely, abnormal estrogen levels can also modulate the gut microbiome and contribute to maintaining the integrity of the intestinal epithelial barrier [ 10 , 11 ]. It is preliminarily hypothesized that disorders of IF and gmGUS are closely related to the development of POF. However, the role of intestinal flora and gmGUS in POF remains largely unexplored. It is unclear whether changes in the composition of gut microbiota and enzymatic activity are associated with ovarian insufficiency, and whether these biomarkers could be used to support early diagnosis or risk stratification in clinical practice. In this study, we investigated the levels of key gut microbiota ( Lactobacillus , Bifidobacterium , Enterococcus , and Escherichia coli ), gmGUS activity, and serum sex hormones, including anti-Müllerian hormone (AMH), follicle-stimulating hormone (FSH), luteinizing hormone (LH), estradiol (E2), in female with POF and healthy controls. We further assessed the correlation between microbial factors and hormone levels and evaluate the diagnostic performance of a combined biomarker model. By addressing this knowledge gap, our study aims to enhance understanding of the gut–ovary axis in POF and explore a novel, non-invasive diagnostic strategy for early detection.

Supplementary Material 1 Supplementary Material 1