This research was a population-based study and was approved by the Medical Ethics Committee of Hainan Women and Children’s Medical Center (Ethics Number: HNWCMC MEC No. 148 of 2025) in China. The study was conducted in strict accordance with the principles of the Declaration of Helsinki. Clinical data and serum samples were collected from 247 women of reproductive age who underwent preconception fertility evaluation at the Hainan Women and Children’s Medical Center from January 2022 to June 2024. The inclusion criteria for all participants were as follows: (1) age between 20 and 35 years; (2) no history of chronic diseases, endocrine disorders, tumors, autoimmune diseases, or active infections; and (3) provision of informed consent to participate in the study. The exclusion criteria were as follows: (1) individuals with incomplete clinical information or missing samples; and (2) those who were pregnant, breastfeeding, or had a history of infertility. Data were collected on the participants’ age, body mass index (BMI), age at menarche, number of pregnancies, menstrual cycle characteristics, income, and education level. Each woman underwent a comprehensive evaluation, which included measurement of FSH, LH, PRL, E₂, P, TT, and anti-müllerian hormone (AMH) during the mornings of days 2 to 4 of her menstrual cycle.

Twenty female C57BL/6J mice, each aged 8 weeks and weighing between 18 and 20 grams, purchased from GemPharmatech Co., Ltd., Nanjing, Jiangsu, China (License No. SCXK(SU)2023-0009), were utilized in this animal study. The sample size was determined based on the degrees of freedom derived from the variance analysis. All procedures and protocols involving mice were conducted in accordance with the guidelines of the Animal Care and Use Committee at Hainan Medical University and were approved by this committee (Ethics Number: HYLL-2024-222). All animals were housed in a controlled environment, with temperature, humidity, and light carefully regulated within a fully staffed, dedicated animal facility operating on a 12-hour light/dark cycle. Veterinarians were available on call at all times. Food and water were provided ad libitum. This research used Python 3.9 (Python Software Foundation, Beaverton, OR, USA) to calculate the required sample size for a one-way Analysis of Variance (ANOVA) power analysis. The anticipated effect size (Cohen’s f) was established at 0.75, with a significance level of $\alpha$ = 0.05 and a statistical power of 0.8, across four groups. Calculations demonstrated that a minimum of four mice per group was required to meet statistical criteria. To account for potential experimental attrition, the study ultimately included five mice per group, resulting in a total of twenty mice. Subsequently, a block randomization procedure utilizing Python 3.9 was implemented to generate a random allocation sequence, evenly distributing twenty mice into four groups of five subjects each. Following a one-week acclimatization period, the mice were randomly allocated into four distinct groups: a control group administered an equal volume of a solution comprising 1% Tween and 2% dimethyl sulfoxide in water; a low-dose PFHpA group at 0.5 mg/kg/day; a medium-dose PFHpA group at 5 mg/kg/day; and a high-dose PFHpA group at 50 mg/kg/day. The doses were established following a comprehensive review of the existing literature [21, 22, 23]. Mice in the treatment cohort received daily oral gavage administrations of PFHpA for 28 consecutive days.

Blood samples were collected from women of reproductive age by venipuncture between days 2 and 4 of their menstrual cycle for preconception fertility evaluation. The samples were then centrifuged at 3000 rpm for 10 minutes at 4 °C and subsequently stored at –80 °C for PFAS analysis.

Methods for measuring PFAS were adapted from existing literature [24, 25]. We utilized ultra-high-performance liquid chromatography-tandem quadrupole mass spectrometry (UHPLC-MS/MS) to analyze the concentrations of 22 different PFAS compounds in 0.1 mL of plasma. To ensure rigorous quality control, each batch of samples included internal quality control samples and procedural blank samples. Specifically, 2 ng of labeled standard and 1 mL of formic acid were added to the plasma samples. After vortexing, the samples were loaded onto the column and sequentially eluted with 1 mL of 0.1 M formic acid, 3 mL of a 50:50 mixture of 0.1 M formic acid and methanol, and 0.5 mL of a 1% ammonium hydroxide solution. PFAS were eluted using 1.8 mL of 1% ammonium hydroxide in acetonitrile. The eluate was then concentrated and diluted to 0.1 mL with a mixture of methanol and 10 mM ammonium acetate before transfer to a UPLC-MS/MS system for analysis. For the current study, only PFAS compounds with a detection rate exceeding 85% in the population were included in the analysis. This encompassed 1 short-chain PFAS (PFHpA), an emerging PFAS substitute (6:2 Cl-PFESA), and 8 legacy PFAS compounds (PFOA, PFOS, perfluorononanoic acid [PFNA], perfluorodecanoic acid [PFDA], PFHxS, perfluoroheptanesulfonic acid [PFHpS], perfluorododecanoic acid [PFDoA], and perfluoroundecanoic acid [PFUndA]). We defined the limit of detection (LOD) as the analyte concentration at which the signal-to-noise ratio (S/N) equals 3. Values below the LOD were substituted with the LOD divided by the square root of two ($\surd$2), which is an operation performed before the logarithmic transformation.

Two days after completing the oral gavage procedure, mice were anesthetized via intraperitoneal injection of a 0.3% sodium pentobarbital solution at a dose of 0.2 mL per 10 g of body weight. The procedure was initiated after the mice were placed under surgical anesthesia. Blood samples were collected via cardiac puncture and left at room temperature for 2 hours. The samples were then centrifuged at 1000 rpm for 20 minutes at 2 °C to 8 °C. The supernatants were then analyzed by enzyme-linked immunosorbent assay (ELISA) to quantify levels of FSH, LH, PRL, E₂, TT, and P. Subsequently, the mice were humanely euthanized via intraperitoneal injection of 3% sodium pentobarbital, administered at a dose of 0.05 mL per 10 grams of body weight.

Total RNA was extracted from mouse ovarian tissues on the same day using the Ultrapure RNA Kit (CW0581M, CWBIO, Taizhou, Jiangsu, China). RNA quality was verified by spectrophotometry, formaldehyde agarose gel electrophoresis, and spectrophotometric analysis. Expression analysis of transcripts related to the steroid pathway was performed using quantitative polymerase chain reaction (qPCR). In brief, cDNA was synthesized from 1 µg of total RNA using the HiScript II Q RT SuperMix (+gDNA wiper) (R223-01, Vazyme, Nanjing, Jiangsu, China) according to the manufacturer’s instructions for qPCR. A PCR reaction mixture containing 1 µL of cDNA was amplified in a total volume of 20 µL using a PCR amplifier (TC-EA, Hangzhou BORI Technology Co., Ltd., Hangzhou, Zhejiang, China), with 0.8 µL of each forward and reverse primer pair (see Table 1). qPCR was performed using a fluorescent real-time PCR instrument (CFX Connect™ Real-Time, Bó Lè Life Medical Products (Shanghai) Co., Ltd., Shanghai, China). Relative fold gene expression changes (2^{-Δ⁢Δ⁢CT}) were calculated in comparison to vehicle controls and normalized to the expression of the reference gene. The transcripts analyzed included CYP11A1, CYP17A1, 3$\beta$-HSD, and 17$\beta$-HSD.

Mean values are expressed as the mean $\pm$ standard deviation (SD), while medians are presented as interquartile ranges (IQR). Categorical data are described using counts and percentages. PFAS concentrations are reported as medians and have been transformed with the natural logarithm to reduce skewness for subsequent analyses. Spearman correlation coefficients were calculated to assess the relationships among PFAS concentrations. Multiple linear regression models were used to examine the association between individual PFAS exposure and reproductive hormones. This study used a hybrid approach that combines prior knowledge with effect size change analysis to identify possible confounding factors. All regression analyses were adjusted for various general characteristics, including age, BMI, age at menarche, number of pregnancies, menstrual cycle irregularity, education, and income level. To manage the risk of false positives from multiple comparisons, we used the Benjamini-Hochberg (BH) method to control for the false discovery rate. Differences between group means were assessed using one-way ANOVA and post hoc tests or non-parametric methods. The significance level for all statistical analyses was established at 0.05 (p 0.05). The analyses were performed utilizing IBM SPSS Statistics version 27 (IBM Corporation, Armonk, NY, USA), with graphical representations created using GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA).

The demographic characteristics of the participants are summarized in Table 2. The mean age was 29.44 years, with a mean BMI of 21.41 kg/m² and a mean age at menarche of 13.15 years. Table 3 presents the median concentrations of reproductive hormones measured in the study, including FSH, LH, PRL, E₂, TT, P, and AMH. The median concentrations were 6.50 IU/L for FSH, 5.61 IU/L for LH, 19.90 ng/mL for PRL, 42.70 pg/mL for E₂, 0.28 ng/mL for TT, 0.22 ng/mL for P, and 3.80 ng/mL for AMH. The median levels of these reproductive hormones were all within the normal range for this age group. As shown in Table 4, the majority of women exhibited detectable levels of all 10 PFAS compounds assessed, with PFOA exhibiting the highest median concentration at 2.295 ng/mL.

Table 5 presents the results of a linear regression analysis examining the associations between individual PFAS compounds and reproductive hormone levels. After adjusting for potential confounders, the natural logarithm of PFHpA was positively correlated with both E₂ and TT levels. Specifically, the $\beta$ coefficient for E₂ was 0.32, with a 95% confidence interval (CI) for B of 12.10 to 29.60 (B = 20.83). For TT, the $\beta$ was 0.51, with a 95% CI for B of 0.18 to 0.30 (B = 0.24). The natural logarithm of PFDA was also positively correlated with TT levels, exhibiting a $\beta$ of 0.32 and a 95% CI for B ranging from 0.01 to 0.40 (B = 0.20). No other PFAS showed significant associations with reproductive hormones.

Fig. 1 illustrates the serum reproductive hormone levels across four mice groups: the control group, the low-dose PFHpA group (0.5 mg/kg/day), the medium-dose PFHpA group (5 mg/kg/day), and the high-dose PFHpA group (50 mg/kg/day). The high-dose PFHpA group exhibited higher TT levels than all other three groups, with all comparisons yielding p 0.0001. Additionally, the control group had lower PRL levels than the three other groups, with all p 0.01. No significant differences were observed for the other reproductive hormones among the four groups.

Fig. 1.

Comparison of serum reproductive hormone levels among the four mice groups. A, Control group; B, Low-dose PFHpA group (0.5 mg/kg/day); C, Medium-dose PFHpA group (5 mg/kg/day); D, High-dose PFHpA group (50 mg/kg/day). ns: p $>$ 0.05; **: p 0.01; ****: p 0.0001.

Fig. 2 shows the relative mRNA expression levels of CYP11A1, CYP17A1, 3$\beta$-HSD, and 17$\beta$-HSD across the four mice groups. Higher relative mRNA expression of CYP11A1 levels were observed in the high-dose PFHpA group compared to the control and low-dose PFHpA groups, with all comparisons yielding p 0.01. Higher CYP11A1 levels were also observed in the medium-dose PFHpA than in the control groups. Furthermore, the high-dose PFHpA group exhibited higher CYP17A1, 3$\beta$-HSD, and 17$\beta$-HSD levels than all other groups, with all comparisons resulting in p 0.05. Significant differences in 3$\beta$-HSD and 17$\beta$-HSD were also observed in the medium-dose PFHpA group compared to the other three groups, with p-values 0.05 for all comparisons.

Fig. 2.

Comparison of CYP11A1, CYP17A1, 3$\beta$-HSD, and 17$\beta$-HSD levels across four mice groups. A, Control group; B, Low-dose PFHpA group (0.5 mg/kg/day); C, Medium-dose PFHpA group (5 mg/kg/day); D, High-dose PFHpA group (50 mg/kg/day). ns: p $>$ 0.05; *: p 0.05; **: p 0.01; ***: p 0.001; ****: p 0.0001.

Our study has several limitations that warrant acknowledgment. First, reproductive hormone levels are known to fluctuate significantly over time, and these variations can markedly reduce the statistical power required to identify significant differences. Furthermore, the doses of PFHpA administered to the mice in this study were significantly higher than those typically encountered by the general population. This discrepancy may bias the evaluation of the causal relationship between PFHpA exposure and sex hormone levels, underscoring the need for further research.