Patients having in vitro fertilization and embryo transfer at the General Hospital of Tianjin Medical University between October 2023 and March 2025 were enrolled in the study. For this study, researchers used the Rotterdam criteria to classify participants into two groups: those with normal ovarian function and those with PCOS. The gonadotropin-releasing hormone antagonist (GnRH-A) protocol was used to induce ovulation in both groups. Patients underwent either fresh embryo transfer or frozen-thawed embryo transfer (FET), with FET protocols including natural cycles, artificial cycles, and stimulated cycles.

Inclusion Criteria for the PCOS Group were based on meeting two of these three factors, after ruling out other causes of elevated androgen levels and ovulatory disorders (1): irregular or absent ovulation (2), clinical or biochemical signs of elevated androgens (3), polycystic ovarian morphology on ultrasound.

Both groups underwent the same ovulation induction protocol, namely GnRH-A protocol. From the second to the third day of menstruation, gonadotropin (Gn) was administered at a dosage of 150–225 IU daily. Ganirelix acetate (0.25 mg/day) was administered subcutaneously starting when the leading follicle achieved ≥14 mm diameter, and continued daily until the day of ovulation trigger (hCG administration). A dosage of 5000–10000 IU of hCG or 250 µg of recombinant hCG was administered after two or three dominant follicles had grown to 18 mm in diameter, and the levels of estradiol per mature follicle varied from 200 to 300 ng/L. A transvaginal aspiration was used to retrieve oocytes 36 to 38 hours following hCG treatment, with the use of ultrasound guidance. On certain days—the day of hCG injection (dhCG), the day of oocyte pickup (dOPU), the day of starting Gn treatment (dGn), and the fifth day of Gn stimulation (dGn5) —blood samples (5 mL) were taken. On the day of oocyte retrieval, follicles measuring 18 mm or larger were harvested for FF ( Figure 1 ).

Prior to further analysis, all samples were spun at 3000 rpm for 10 minutes. The resulting supernatant was then kept at -80 °C in 1.5 mL EP tubes. In order to quantify AMH and INHB in serum and FF, ELISA kits were utilized.

The samples were retrieved from the -80 °C freezer and thawed in a 4 °C refrigerator until completely melted before proceeding with the assay. Prior to use, all reagents from the ELISA kits were allowed to reach room temperature (18-25 °C). The ELISA kits were employed to detect human INHB (Cat. No. E-EL-H0313, Elabscience) and human AMH (Cat. No. E-OSEL-H0004, Elabscience).

Clinical data were retrospectively extracted from the Electronic Medical Record system for patients diagnosed with PCOS undergoing the IVF/ICSI cycle. Key indicators included: PCOS diagnostic markers (Basal testosterone (T, ng/ml) measured on cycle days 2-4; Antral Follicle Count (AFC) assessed via transvaginal ultrasound on days 2-4); COH treatment parameters (Total gonadotropins dosage (IU); Duration of COH (days); Serum E2 level on trigger day (pg/ml)); and Embryology laboratory outcomes (Number of retrieved oocytes; 2PN fertilizations; Available embryos; Top-quality Embryo (TQE)). TQE was specifically defined according to the Istanbul Consensus as top-quality cleavage-stage embryos on Day 3: possessing 8 cells (or 7–9 cells), ≤10% fragmentation, uniform blastomere size, and absence of multinucleation (18). TQE rate (%) was calculated accordingly. Data extraction was performed by two independent clinicians using a dual-blinded approach with cross-verification for accuracy.

Statistical analysis was performed using GraphPad Prism 9.0 software. During the data preprocessing phase, all variables were initially subjected to normality testing using the Shapiro-Wilk test. Normally distributed quantitative data were expressed as mean ± standard deviation (Mean ± SD), with intergroup comparisons conducted using independent samples t-test (for two groups) or one-way ANOVA (for multiple groups) according to sample size characteristics. Conversely, non-normally distributed quantitative data were presented as median (interquartile range) [M(QR)], with between-group differences analyzed through Mann-Whitney U test (for two groups) or Kruskal-Wallis H test (for multiple groups). Categorical variables were described using frequency (percentage). The linear correlation between variables was assessed using Pearson correlation coefficient, while the association analysis of binary variables was performed through receiver operating characteristic (ROC) curve analysis. For all statistical tests, a two-tailed P-value <0.05 was considered the threshold for determining statistical significance.

The research was approved by the Ethics Committee of Tianjin Medical University General Hospital before it was conducted. After being apprised of their rights, possible advantages, and hazards, all participants were asked to sign informed consent forms (Ethics: IRB2023-KY-332).

Forty patients with PCOS and forty controls were part of an 80-person IVF cohort ( Table 1 ). There were substantial differences in the baseline levels of testosterone and luteinizing hormone, with the former having significantly higher levels (P<0.05, P<0.01). Furthermore, there was a significant increase in the number of antral follicles in the group with PCOS (P<0.0001). Among the other baseline variables, there were no significant differences between the groups (P>0.05). Remarkably, there were differences in the PCOS group with regard to the number of days of gonadotropin administration needed, E2 levels on the trigger day, number of oocytes recovered, two pronuclei (2PN) fertilization and available embryos (P<0.05, P<0.0001, P<0.0001, P<0.0001, P<0.001). Surprisingly, the PCOS group had a lower rate of top-quality embryos (TQE) (P<0.0001).

Research indicates that serum AMH concentrations in the PCOS cohort were elevated throughout the COH, showing a marked statistical difference (P<0.0001). Notably, serum AMH levels displayed a declining trend in both groups. When comparing the serum AMH levels of dGn5, dhCG, and dOPU with baseline levels within each group, it was observed that in the PCOS cohort, levels at dhCG and dOPU were reduced compared to dGn, with significant differences noted (P<0.0001, P<0.0001). A similar pattern was observed in the control group (P<0.001, P<0.0001). However, on days dGn5, dhCG, and dOPU, INHB levels in the PCOS group, reaching significance (P<0.0001, P = 0.0001, P<0.01). The trajectory of serum INHB levels increased from dGn to dhCG and subsequently decreased at dOPU. Comparative analyses within each group from baseline levels on dGn5, dhCG, and dOPU revealed that INHB levels in the PCOS group were higher on dGn5, dhCG and dOPU compared to dGn, with significant differences (P<0.01, P<0.0001, P<0.01). In the control group, INHB levels on dhCG and dOPU exceeded those on dGn, marking significant differences (P <0.0001, P<0.001). On the day of oocyte pickup, FF AMH levels were significantly higher in the PCOS group (P=0.0002). This was also true for INHB levels, which were elevated in the PCOS group (P<0.05) ( Table 2 ).

Figure analysis illustrates correlations between AMH and INHB levels at several time points (dGn5, dhCG, dOPU) and in FF. On dGn5, a positive correlation is noted between AMH and INHB levels (P<0.05), suggesting that increases in AMH levels are generally accompanied by rises in INHB levels, though this association is not particularly robust (R² = 0.0688). At dhCG, the correlation strength between AMH and INHB levels intensifies (P<0.0001), with data demonstrating a more pronounced upward trajectory, indicating that rises in AMH are likely paralleled by significant increases in INHB levels (R² = 0.2126). At dOPU, the relationship persists (P<0.05), with the strength of correlation remaining moderate (R² = 0.0628). In FF, a slight negative correlation emerges between AMH and INHB levels (P < 0.05), where increases in AMH levels correspond with minor decreases in INHB levels (R² = 0.0603). Overall, the correlation strength varies across different time points and in FF, with dhCG demonstrating the strongest positive correlation ( Figure 2 ).

This study employed Pearson linear correlation analysis to evaluate the associations between serum and follicular fluid AMH levels and PCOS diagnostic markers and COH outcomes. Serum AMH levels at dGn, dGn5, dhCG, and dOPU showed positive correlations with AFC, trigger-day E2 levels, number of oocytes retrieved, number of 2 pronuclei (2PN) embryos, and number of usable embryos, while demonstrating negative correlations with TQE rate. Additionally, AMH levels at dGn5, dhCG, and dOPU showed positive correlations with baseline T. However, AMH at dGn showed no significant correlation with baseline T. Moreover, serum AMH levels at dGn, dhCG, and dOPU were positively correlated with the number of TQE. No statistically significant correlations were observed between serum AMH levels and Gn dosage or duration of COH. Follicular fluid AMH levels exhibited a positive correlation with AFC, while demonstrating a negative correlation with the TQE rate, however, no significant associations were observed with Gn dosage, COH duration, trigger-day E₂ levels, number of oocytes retrieved, 2PN parameters, number of usable embryos, or number of TQE. (detailed correlation coefficients and p-values are presented in Table 3 ).

This study employed Pearson linear correlation analysis to evaluate the associations between serum and follicular fluid INHB levels and PCOS diagnostic markers and COH outcomes. On dGn5, serum INHB concentrations exhibited a positive correlation with AFC, E2 levels on the trigger day and the number of oocytes collected, while demonstrating a negative correlation with the TQE rate. At dhCG, INHB concentrations exhibited a positive correlation with AFC, E2 levels on the trigger day, the quantity of oocytes retrieved, the incidence of 2PN fertilizations, and the count of viable embryos. In the context of dOPU, there was a positive correlation observed between INHB concentrations and both COH duration and E2 levels on the trigger day. Conversely, a negative correlation was noted with the TQE rate. No significant correlations were found between serum INHB levels on dGn and those in FF, nor with Gn dosage, COH duration, E2 levels on the trigger day, the number of oocytes retrieved, the number of 2PN fertilizations, the count of viable embryos, or the TQE rate (detailed correlation coefficients and p-values are presented in Table 4 ).

The outcome of the multiple linear regression analysis indicated that the regression model was significant. Specifically, after adjustment for age, BMI, and baseline LH, AMH levels on dOPU (β= -5.2250, P = 0.0014) and INHB levels on dOPU (β=-0.1106, P = 0.0252) remained significant negative predictors of the TQE rate. The adjusted model explained 22.6% of the variance in the rate of TQE. These results demonstrate that the dual-hormone prediction model incorporating both AMH and INHB exhibits superior predictive performance compared to single-hormone models ( Table 5 ).

This study continued to track the pregnancy outcomes of patients in both groups and collected data to compare differences in cumulative pregnancy rates. Cumulative pregnancy rate is the probability of achieving at least one clinical pregnancy after transferring all available embryos (fresh + frozen) derived from a single oocyte retrieval cycle. The cumulative pregnancy rate was 47.50% (19/40) in the PCOS group and 77.50% (31/40) in the control group. Chi-square test results indicated a statistically significant difference in cumulative pregnancy rates between the two groups (χ²=7.68, P = 0.0056), with the PCOS group demonstrating a significantly lower cumulative pregnancy rate (47.50%) compared to the control group (77.50%) ( Table 6 ).

Building upon these findings, the study continued to track pregnancy outcomes until live birth. As of March 2025, the PCOS group had undergone 71 embryo transfer cycles, resulting in 7 live births. The control group had 66 embryo transfer cycles with 18 live births. The live birth rate was calculated using the formula: live birth rate = (number of live birth cycles/number of embryo transfer cycles)×100. This yielded a live birth rate of 9.86% (7/71) in the PCOS group and 27.27% (18/66) in the control group. Chi-square analysis revealed a statistically significant difference in live birth rates between the two groups (χ²=6.95, P = 0.0084), with the PCOS group demonstrating a significantly lower live birth rate (9.86%) compared to the control group (27.27%) ( Table 7 ).

Following the completion of pregnancy outcome data collection in the two study groups, this research further explored the association between AMH, INHB, and clinical pregnancy outcomes. To investigate this association, we performed ROC curve analysis to evaluate the predictive capacity of these hormones for pregnancy outcomes. The ROC analysis revealed significant differences in the predictive efficacy of serum hormone levels and follicular fluid biomarkers at different timepoints. Specifically, serum INHB at dGn5 (day 5 of gonadotropin stimulation) demonstrated predictive value for pregnancy outcomes (AUC = 0.71, 95% confidence interval: 0.59–0.83). The optimal cutoff value for serum INHB at dGn5 was determined to be 29.15 pg/mL, with a sensitivity of 0.63 and specificity of 0.71. In contrast, neither serum AMH at any timepoint nor follicular fluid AMH and INHB levels showed significant predictive power, as their ROC curves did not exceed random prediction thresholds (AUC<0.65, P>0.05) ( Figure 3 ). These findings suggest that serum INHB (dGn5) may serve as a potential biomarker for individualized prediction of pregnancy outcomes in PCOS patients.