Elevated maternal pre-transfer serum lipid peroxidation is associated with implantation failure and early pregnancy loss.

This study included 2527 patients undergoing IVF/ICSI cycles at the Reproductive Hospital Affiliated with Shandong University (March 2023–May 2024). Serum samples were collected from 2040 patients 1 day before the scheduled fresh embryo transfer, and intrauterine fluid samples were obtained from 487 patients on the embryo transfer day. Inclusion criteria comprised women aged 20–40 years undergoing embryo transfer with either pre-transfer serum collection or intrauterine fluid collection, and complete clinical data. Exclusion criteria encompassed contraindications to ART/pregnancy (e.g., uncontrolled hypertension/diabetes), autoimmune diseases, abnormal liver/kidney function, history of malignancy, endometrial thickness < 7 mm pre-transfer, severe intrauterine adhesions, known genetic factors, current ectopic pregnancy, active gynecological infections, or significant missing data. Participants were first classified according to established criteria for RIF, with the RIF group comprising individuals who failed to achieve a clinical pregnancy after the cumulative transfer of ≥ 4 good-quality embryos across ≥ 3 cycles, and all other participants designated as non-RIF [ 31 ]. To examine the association between maternal lipid peroxidation and early pregnancy outcomes, patients were prospectively followed during their fresh embryo transfer cycle and categorized by outcome. The successful pregnancy group included patients with ongoing intrauterine pregnancy who were confirmed to have viable fetuses with detectable fetal cardiac activity at 12 weeks of gestation. The failed pregnancy group comprised patients experiencing either implantation failure or early pregnancy loss, specifically defined by the following conditions: serum β-hCG < 5 mIU/mL 14 days post-transfer, β-hCG ≥ 5 mIU/mL with no subsequent gestational sac on ultrasound, or spontaneous miscarriage before 12 weeks post-confirmation of gestational sac. This prospective cohort study was conducted at the Reproductive Hospital of Shandong University. Patient cohorts were established through an initial retrospective screening of the hospital’s electronic medical records. Patients meeting the predefined diagnostic criteria for recurrent implantation failure (RIF) were enrolled to form the RIF cohort, while a control cohort consisted of patients without RIF. Lipid peroxidation markers were quantified and compared between these groups. To investigate the association between pre-embryo transfer lipid peroxidation levels and subsequent early pregnancy outcomes, all enrolled participants underwent prospective follow-up from March 2023 to August 2024 according to a standardized protocol. This translational study integrated clinical data analysis with a complementary murine model. Clinical data from an IVF/ICSI cohort were analyzed using propensity score matching (caliper = 0.02) to minimize confounding. Patients were first stratified into those with or without RIF for comparative analysis of pre-transfer serum biomarkers, using logistic regression to identify factors associated with RIF. Subsequently, patients without RIF were reclassified based on early pregnancy outcomes into a successful pregnancy group and a failed pregnancy group. Associations between biomarker levels and outcomes were analyzed using RCS regression (4 knots; model adjusted for age, BMI, number of embryos transferred, and endometrial thickness) and subgroup analyses. Intrauterine MDA levels during transfer were compared between these outcome groups. In parallel, a murine model of peri-implantation lipid peroxidation was established by administering erastin to evaluate its association with endometrial receptivity and decidualization. Wild-type C57BL/6 mice (Jicui Lab Animal Co., Ltd.) were housed under specific pathogen-free (SPF) conditions. Following confirmation of a vaginal plug (gestational day 1, GD1), mice received intraperitoneal injections of either erastin (20 mg/kg) or vehicle control on GD2 and GD3, based on prior methodology [ 32 – 34 ]. Serum and tissues were collected on GD4 for analysis (Fig. S2a). The erastin solution was prepared by dissolving erastin (Selleck, S7242) in DMSO (Solarbio, D8371) to generate a 5 mg/mL stock, which was then mixed with polyethylene glycol 300 (PEG300) (Selleck, S6704), Tween 80 (Selleck, S6702), and 0.9% NaCl to form the working injection solution. The vehicle control was prepared identically by replacing the erastin stock with plain DMSO. Fasting venous blood (5 mL) was collected pre-transfer, stored at 4 °C immediately, and processed within 2 h. Serum was separated by centrifugation (12,000 × g, 10 min), aliquoted into three portions, and stored at − 80 °C. Samples were equilibrated to room temperature pre-analysis. Uterine cavity fluid was collected post-transfer via catheter flushing with 500 μL PBS, vacuum-concentrated to 100 μL for MDA quantification. All procedures followed strict sterile technique. Mouse serum was collected by retro-orbital bleeding under isoflurane anesthesia, centrifuged (12,000 × g, 10 min, 4 °C), and stored at − 80 °C. Female mice were mated overnight, and the presence of a vaginal plug the following morning was designated GD1. Uterine tissues were collected from GD4 to GD8. Successful pregnancy was confirmed on GD4 by flushing embryos from the uterine lumen. On GD5 and GD6, embryo implantation sites were visualized using Chicago blue dye staining. Liver, kidneys, and small intestine were concurrently harvested. Frozen Sects. (12 μm thick) were prepared at − 20 °C in OCT compound and stored at − 80 °C. For immunofluorescence (IF) and immunohistochemistry (IHC), sections were thawed, air-dried, fixed in 4% PFA, and blocked before incubation with primary antibody (1:300) overnight at 4 °C. After washing, sections were incubated with secondary antibody (1:500): for IF, this included DAPI (1:100) and proceeded for 1 h at room temperature before mounting and imaging; for IHC, incubation lasted 2 h at 37 °C, followed by DAB staining, hematoxylin counterstaining, dehydration, clearing, and mounting. All images were acquired at 20 × magnification, and antibody details are provided in Additional file 1: Table S1. Frozen sections were equilibrated and hybridized with digoxigenin (DIG)-labeled probes ( Ltf , Bmp2 , Prl8a2 , Prl3c1 , Hoxa10 , Wnt4 ), which were synthesized using the SP6/T7 transcription kit (Roche, 11,175,025,910) with DIG-RNA Labeling Mix (Roche, 11,277,073,910), followed by DNase I (New England Biolabs, M0303S) digestion and ethanol precipitation. Sections were fixed in 4% PFAT, processed through methanol gradients, digested with proteinase K, and acetylated. After hybridization, slides were blocked, incubated with anti-DIG-AP antibody (Roche, 11,093,274,910), and signals were developed with BCIP/NBT (Roche, 11,697,471,001) for bright-field microscopy (Olympus, VS120). Probe sequences are listed in Additional file 1: Table S2. Total RNA was isolated from 100 mg of uterine tissue using TRIzol reagent (Invitrogen, 15,596–018). Following homogenization and phase separation with chloroform, RNA was precipitated with isopropanol, washed with 75% ethanol, and dissolved in RNase-free water. RNA quality was assessed using a NanoDrop OneC spectrophotometer (Thermo Fisher Scientific) and agarose gel electrophoresis. cDNA was synthesized from 1 μg of total RNA using the Evo M-MLV Reverse Transcription Kit (Absin, AG11711) after genomic DNA removal with gDNA Eraser (Absin, AG11711). Quantitative PCR was performed using SYBR Green Master Mix (Vazyme, Q311-02) on a Light Cycler 96 system, with expression levels normalized to Rpl7 and calculated via the ΔCt method. Primer sequences are listed in Additional file 1: Table S3. Serum of mice was collected on the 4th day of pregnancy. The levels of P4 and E2 were measured via enzyme-linked immunosorbent assay (ELISA) kits (Cayman, 582,601 and 501,890, respectively). MDA and non-heme iron levels were measured as described previously [ 35 – 37 ]. Serum MDA was quantified using a thiobarbituric acid-reactive substances assay with 1-methyl-2-phenylindole (Shanghai Yuanye Biotechnology, S46375 ), and absorbance was read at 586 nm. Non-heme iron was determined colorimetrically using 4,7-diphenyl-1,10-phenanthroline sulfonate sodium salt (Solarbio, 5,274,649–3) with absorbance measured at 535 nm. 12-HETE and 15-HETE were quantified using commercial ELISA kits (Abcam, ab133034 and ab133035), and T-AOC and SOD activity were assessed using kits from Nanjing Jiancheng (A015-2–1, A001-3–2). Absorbance for plate-based assays was measured on a TECAN INFINITE 200 PRO microplate reader. For tissue analyses, homogenates were prepared at a 1:10 (w/v) ratio in lysis buffer, and protein concentrations were normalized using a bicinchoninic acid (BCA) assay prior to analysis. Cellular lipid peroxidation in GD4 murine endometrial stromal cells was assessed by flow cytometry using BODIPY™ 581/591 C11 staining (1:1000), with data analyzed in FlowJo™ 10. All reagents were from a single batch, and inter-plate consistency was controlled. Data analysis utilized SPSS 26.0 for basic statistics, GraphPad Prism 9.5 for boxplots, and R 4.4.0 for RCS, subgroup analyses, and forest plots. Normality was assessed via Shapiro–Wilk tests and histograms. Continuous data were presented as mean ± SD (normally distributed) or median [IQR] (nonnormal), and were compared using independent t -tests or Mann–Whitney U tests, respectively. Categorical variables were expressed as n (%) and were analyzed by χ 2 or Fisher’s exact tests with Bonferroni correction. Missing data were handled via multiple imputation. Univariate and multivariate logistic regression identified predictors of pregnancy outcomes. To mitigate scale effects, non-heme iron (converted from μg/dL to μg/mL [× 0.01]), 12-HETE, and 15-HETE (pg/mL to ng/mL × 0.001) were standardized. Standardized variables underwent Kolmogorov–Smirnov testing, with multicollinearity assessed by variance inflation factors (VIF < 5). Results are reported as adjusted odds ratios (ORs) with 95% confidence intervals. All tests were two-sided ( α  = 0.05).

A total of 2040 patients were included (306 with RIF and 1734 without RIF). Significant baseline differences were observed in age, AFC, AMH levels, number of oocytes retrieved, and number of failed embryo transfer cycles (Additional file 1: Table S4). Following 1:4 propensity score matching (PSM), a greater number of failed embryo transfer cycles was the only characteristic that remained significantly different between the two matched groups (Table  1 ). Table 1 Comparison of clinical baseline characteristics, lipid peroxidation, and antioxidant levels between RIF and non-RIF groups Variable Non-RIF group ( n  = 1204) RIF group ( n  = 302) P value Woman age, year 33.61 (3.53) 33.73 (3.53) 0.61 BMI, kg/m 2 23.56 (3.54) 23.42 (3.67) 0.54 No. of AFC 8.67 (4.86) 8.54 (4.90) 0.70 No. of retrieved oocytes 8 (5, 12) 9 (5, 13) 0.59 No. of failed embryo transfer cycle 0 (0, 1) 3 (3, 4)  < 0.001 Baseline FSH, IU/L 7.34 (2.73) 7.11 (2.90) 0.18 TSH, μIU/mL 2.28 (1.46) 2.22 (1.02) 0.51 AMH, ng/mL 2.50 (1.36, 4.16) 2.46 (1.31, 4.12) 0.89 HCG-estradiol, pg/mL 2435.35 (1603.20) 2372.74 (1556.84) 0.54 HCG-progesterone, ng/mL 0.61 (0.34) 0.60 (0.35) 0.51 MDA, μM 5.58 (2.49) 6.27 (2.77)  < 0.001 Non-heme iron, μg/dL 337.45 (202.36) 393.36 (189.91)  < 0.001 T-AOC, μM 0.86 (0.15) 0.86 (0.14) 0.70 SOD, U/mL 41.68 (5.85) 40.76 (6.64) 0.03 Data are presented as mean (standard deviation) and median (interquartile range). P  < 0.05 was considered statistically significant. Comparison of clinical baseline characteristics, lipid peroxidation, and antioxidant levels between RIF and non-RIF groups Data are presented as mean (standard deviation) and median (interquartile range). P  < 0.05 was considered statistically significant. Serum analysis showed that patients with RIF had elevated levels of MDA and non-heme iron, along with reduced SOD activity, compared to those without RIF ( P   0.05). Univariate regression analysis indicated that elevated levels of MDA (OR = 1.10, 95% CI: 1.053–1.157; P  < 0.001) and non-heme iron (OR = 1.13, 95% CI: 1.067–1.196; P  < 0.001) were associated with RIF. These associations remained significant after adjusting for age, BMI, AMH, TSH, and AFC: adjusted ORs were 1.10 (95% CI: 1.049–1.154; P  < 0.001) for MDA and 1.12 (95% CI: 1.061–1.191; P  < 0.001) for non-heme iron (Fig.  1 e). These results indicate that patients with RIF exhibit markedly higher lipid peroxidation levels and that elevated MDA and non-heme iron are independently associated with RIF. Fig. 1 Elevated serum lipid peroxidation levels are associated with RIF. a – d Serum levels of MDA, non-heme iron, SOD, and T-AOC in PSM cohorts (RIF: n  = 302; non-RIF: n  = 1204). Data presented as mean ± SEM. Boxplots depict medians (midline), interquartile ranges (box boundaries), and 1.5 × IQR whiskers; outliers shown as points. e Univariate and multivariate logistic regression of lipid peroxidation/antioxidant markers with RIF. * P  < 0.05, *** P  < 0.001; ns, not significant Elevated serum lipid peroxidation levels are associated with RIF. a – d Serum levels of MDA, non-heme iron, SOD, and T-AOC in PSM cohorts (RIF: n  = 302; non-RIF: n  = 1204). Data presented as mean ± SEM. Boxplots depict medians (midline), interquartile ranges (box boundaries), and 1.5 × IQR whiskers; outliers shown as points. e Univariate and multivariate logistic regression of lipid peroxidation/antioxidant markers with RIF. * P  < 0.05, *** P  < 0.001; ns, not significant This prospective study evaluated the association between pre-transfer serum markers of lipid peroxidation and antioxidant status with early pregnancy outcomes. Among patients without RIF, significant baseline differences were observed between the successful pregnancy group and the failed pregnancy group (Additional file 1: Table S5). After 1:1 PSM ( n  = 497 per group), these characteristics were balanced (Table  2 ). In the matched non-RIF cohort, the failed pregnancy group had significantly higher serum levels of lipid peroxidation markers (MDA, non-heme iron, 12-HETE, 15-HETE; P  < 0.05) and lower SOD activity ( P   0.05; Table  2 , Fig.  2 a–f). Table 2 Comparison of clinical baseline characteristics, lipid peroxidation, and antioxidant levels between the successful and failed pregnancy groups in patients without RIF Variable Patients without RIF Successful group ( n  = 497) Failed group ( n  = 497) P value Woman age, year 32.89 (3.75) 33.20 (3.84) 0.21 BMI, kg/m 2 23.32 (3.72) 23.64 (3.39) 0.15 No. of AFC 8.49 (3.88) 8.06 (4.02) 0.08 No. of retrieved oocytes 8.97 (4.22) 8.50 (4.47) 0.09 Infertility type  Primary infertility 269/497 (54.1%) 241/497 (48.5%) 0.08  Secondary infertility 228/497 (45.9%) 256/497 (51.5%) Duration of infertility, year 3.00 (2.00, 5.00) 3.00 (1.50, 5.00) 0.60 No. of failed embryo transfer cycle 0 (0, 0) 1 (1, 1)  < 0.001 Baseline FSH, IU/L 7.28 (2.42) 7.45 (2.61) 0.29 TSH, μIU/mL 2.25 (1.08) 2.28 (1.14) 0.65 AMH, ng/mL 2.96 (2.32) 2.82 (2.12) 0.33 HCG-estradiol, pg/mL 2258.74 (1205.72) 2221.86 (1196.15) 0.63 HCG-progesterone, ng/mL 0.58 (0.26) 0.61 (0.28) 0.07 Endometrial thickness, cm 1.04 (0.17) 1.03 (0.15) 0.19 MDA, μM 5.26 (2.44) 5.89 (2.50)  < 0.001 Non-heme iron, μg/dL 318.83 (170.11) 375.02 (224.42)  < 0.001 12-HETE, pg/mL 18,081.78 (16,779.96) 21,881.07 (19,448.04) 0.001 15-HETE, pg/mL 2263.50 (1553.45) 2961.86 (2242.96)  < 0.001 T-AOC, μM 0.87 (0.17) 0.86 (0.14) 0.67 SOD, U/mL 42.23 (6.29) 40.77 (5.91)  < 0.001 Data are presented as the mean (standard deviation), median (interquartile range), or n (%). P  < 0.05 was considered statistically significant. Fig. 2 Comparison of serum lipid peroxidation and antioxidant levels between the successful and failed pregnancy groups in patients without RIF. a – f Serum levels of MDA, non-heme iron, 12-HETE, 15-HETE, SOD, and T-AOC in PSM cohorts (successful pregnancy: n  = 497; failed pregnancy: n  = 497). Data presented as mean ± SEM. Boxplots depict medians (midline), interquartile ranges (box boundaries), and 1.5 × IQR whiskers; outliers shown as points. ** P  < 0.01, *** P  < 0.001; ns, not significant Comparison of clinical baseline characteristics, lipid peroxidation, and antioxidant levels between the successful and failed pregnancy groups in patients without RIF Data are presented as the mean (standard deviation), median (interquartile range), or n (%). P  < 0.05 was considered statistically significant. Comparison of serum lipid peroxidation and antioxidant levels between the successful and failed pregnancy groups in patients without RIF. a – f Serum levels of MDA, non-heme iron, 12-HETE, 15-HETE, SOD, and T-AOC in PSM cohorts (successful pregnancy: n  = 497; failed pregnancy: n  = 497). Data presented as mean ± SEM. Boxplots depict medians (midline), interquartile ranges (box boundaries), and 1.5 × IQR whiskers; outliers shown as points. ** P  < 0.01, *** P  < 0.001; ns, not significant In contrast, patients with RIF exhibited no significant baseline differences between groups with different pregnancy outcomes (Additional file 1: Table S6). In this cohort, the successful pregnancy group had lower serum MDA ( P  < 0.05) and higher T-AOC ( P  < 0.05) compared to the failed pregnancy group (Additional file 1: Table S6; Additional file 2: Fig. S1a–d). Given the limited number of successful pregnancies in the RIF patients, validation in larger cohorts is warranted. Given the substantially larger sample size of the non-RIF cohort, we focused our logistic regression analysis on this group to evaluate the association between pre-transfer lipid peroxidation markers and early pregnancy outcomes. The model defined pregnancy failure at 12 weeks of gestation as the outcome variable, including those with implantation failure and early pregnancy loss. Multivariate logistic regression, adjusted for covariates, identified MDA (OR = 1.11, 95% CI: 1.055–1.170, P  < 0.001), non-heme iron (OR = 1.17, 95% CI: 1.091–1.255, P  < 0.001), 12-HETE (OR = 1.01, 95% CI: 1.004–1.019, P  = 0.002), 15-HETE (OR = 1.24, 95% CI: 1.148–1.341, P  < 0.001), and SOD (OR = 0.96, 95% CI: 0.940–0.981, P  < 0.001) as significant independent factors associated with the outcome (Fig.  3 a). Elevated levels of MDA, non-heme iron, 12-HETE, and 15-HETE were significantly associated with an increased likelihood of pregnancy failure. In contrast, higher SOD levels were inversely associated with the outcome (Fig.  3 a). Fig. 3 Elevated serum lipid peroxidation levels are associated with pregnancy failure in patients without RIF. a Logistic regression of lipid peroxidation/antioxidant markers with early pregnancy outcomes in patients without RIF. b – g RCS analyses for MDA, non-heme iron, 12-HETE, 15-HETE, SOD, and T-AOC. Dashed horizontal line: OR = 1; vertical dashed lines: reference concentrations at OR = 1. Blue shading: 95% CI Elevated serum lipid peroxidation levels are associated with pregnancy failure in patients without RIF. a Logistic regression of lipid peroxidation/antioxidant markers with early pregnancy outcomes in patients without RIF. b – g RCS analyses for MDA, non-heme iron, 12-HETE, 15-HETE, SOD, and T-AOC. Dashed horizontal line: OR = 1; vertical dashed lines: reference concentrations at OR = 1. Blue shading: 95% CI To characterize the dose–response relationships, we performed RCS analyses. In models adjusted for age, BMI, number of embryos transferred, and endometrial thickness (Fig.  3 b–g), MDA, non-heme iron, 12-HETE, and T-AOC demonstrated significant nonlinear relationships with the odds of pregnancy failure ( P -nonlinear  0.05). Collectively, these RCS findings refine the associations, showing that specific lipid peroxidation markers and antioxidants relate to the odds of pregnancy failure through distinct linear and nonlinear patterns. Based on established evidence linking maternal age, BMI, PCOS, and endometriosis to excessive ROS generation and lipid peroxidation [ 38 – 40 ], we conducted subgroup analyses to examine the association of pre-transfer lipid peroxidation and antioxidant markers with early pregnancy outcomes in patients without RIF. When stratified by age (< 35 vs. ≥ 35 years), BMI (< 28 vs. ≥ 28 kg/m 2 ), and the presence of PCOS or endometriosis, the associations of lipid peroxidation markers (MDA, non-heme iron, 12-HETE, 15-HETE, SOD) with the outcome varied across subgroups. Significant associations ( P  < 0.05) were most evident in younger patients (< 35 years) with BMI < 28 kg/m 2 and without PCOS or endometriosis, persisting in both unadjusted and adjusted models. In contrast, no significant associations were observed in patients with BMI ≥ 28 kg/m 2 or those with PCOS. Weaker or non-significant associations were noted in the advanced maternal age (≥ 35 years) and endometriosis subgroups: the association with MDA was not statistically significant in the ≥ 35-year group, while in patients with endometriosis, the point estimates for MDA and non-heme iron were attenuated and non-significant. These results suggest that the associations between lipid peroxidation markers and early pregnancy outcomes may be attenuated or absent in patients with advanced age, obesity, PCOS, or endometriosis (Fig.  4 ). Fig. 4 Subgroup forest plot of the association between lipid peroxidation levels and early pregnancy outcomes in patients without RIF. Subgroup analyses of univariate and multivariate logistic regression for MDA, non-heme iron, 12-HETE, 15-HETE, SOD, and T-AOC with early pregnancy outcomes in patients without RIF. Adjusted according to age, BMI, number of retrieved oocytes, and endometrial thickness. N , number of patients. OR, odds ratio; CI, confidence interval. P  < 0.05 was considered statistically significant Subgroup forest plot of the association between lipid peroxidation levels and early pregnancy outcomes in patients without RIF. Subgroup analyses of univariate and multivariate logistic regression for MDA, non-heme iron, 12-HETE, 15-HETE, SOD, and T-AOC with early pregnancy outcomes in patients without RIF. Adjusted according to age, BMI, number of retrieved oocytes, and endometrial thickness. N , number of patients. OR, odds ratio; CI, confidence interval. P  < 0.05 was considered statistically significant While systemic serum lipid peroxidation reflects maternal status, local intrauterine lipid peroxidation may be directly related to embryo implantation and early pregnancy development. To investigate this, maternal uterine fluid MDA levels, measured on the day of embryo transfer as an indicator of intrauterine lipid peroxidation, were compared between patients with ongoing pregnancy and those with pregnancy failure. In the initial cohort of 487 patients, significant intergroup differences in age, AMH, and number of failed embryo transfer cycles were observed (Additional file 1: Table S7). Following 1:1 PSM to reduce confounding, 215 matched pairs demonstrated balanced baseline characteristics (Additional file 1: Table S8). Notably, intrauterine MDA levels were significantly higher in patients experiencing pregnancy failure than in those with ongoing pregnancy ( P  < 0.05; Additional file 1: Table S8, Fig.  5 a). This difference suggests that elevated intrauterine MDA is associated with increased odds of pregnancy failure. RCS analysis further confirmed a statistically significant nonlinear association, with a trend of decreasing probability of ongoing pregnancy observed at higher intrauterine MDA levels (Fig.  5 b). Fig. 5 Elevated uterine fluid MDA levels are associated with increased odds of pregnancy failure in IVF/ICSI cycles. a Uterine fluid MDA levels in 1:1 PSM cohorts (successful pregnancy: n  = 215; failed pregnancy: n  = 215). Data presented as mean ± SEM. Boxplots depict medians, interquartile ranges, and 1.5 × IQR whiskers; outliers shown as points. *** P  < 0.001. b RCS regression of MDA with pregnancy outcomes. Dashed horizontal line: OR = 1; vertical dashed line: reference concentration at OR = 1. Blue shading: 95% CI Elevated uterine fluid MDA levels are associated with increased odds of pregnancy failure in IVF/ICSI cycles. a Uterine fluid MDA levels in 1:1 PSM cohorts (successful pregnancy: n  = 215; failed pregnancy: n  = 215). Data presented as mean ± SEM. Boxplots depict medians, interquartile ranges, and 1.5 × IQR whiskers; outliers shown as points. *** P  < 0.001. b RCS regression of MDA with pregnancy outcomes. Dashed horizontal line: OR = 1; vertical dashed line: reference concentration at OR = 1. Blue shading: 95% CI Although our clinical data indicated that elevated maternal lipid peroxidation was associated with pregnancy failure, whether increased lipid peroxidation directly impairs endometrial function remained unclear. To explore this possibility and to mimic the elevated maternal lipid peroxidation observed in patients during the peri-implantation period, we established a murine lipid peroxidation model by administering a classical lipid peroxidation inducer during the pre-implantation stage (gestational days 2–3, GD2–GD3) in mice (Additional file 2: Fig. S2a). On GD4, erastin-treated mice exhibited significantly elevated serum MDA and non-heme iron versus controls ( P  < 0.05), while T-AOC and SOD remained unchanged (Additional file 2: Fig. S2b). Uterine MDA was significantly increased in erastin-treated mice (Additional file 2: Fig. S3a, b), unlike liver, small intestine, or ovaries (Additional file 2: Fig. S3c–e). Uterine qRT-PCR revealed significant upregulation of Ptgs2 , Ftl , and Fth mRNA in erastin-treated mice (Additional file 2: Fig. S4a). Immunohistochemistry confirmed reduced GPX4 protein expression (Additional file 2: Fig. S4b) with unchanged ACSL4 (Additional file 2: Fig. S4c). Flow cytometry using BODIPY™ 581/591 C11 demonstrated significantly elevated lipid peroxidation in endometrial stromal cells from treated mice (Additional file 2: Fig. S4d), consistent across three replicates (Additional file 2: Fig. S4e). Erastin administration during GD2 and GD3 resulted in elevated maternal and uterine lipid peroxidation during the implantation window, providing a murine model relevant to the human peri-transfer condition. Lipid peroxidation significantly impaired embryonic implantation and early pregnancy outcomes. Compared with vehicle controls, erastin-treated mice exhibited reduced litter size (Fig.  6 a), fewer embryo implantation sites on GD 5 (Chicago blue dye; Fig.  6 b, c), and abnormal implantation sites with blurred boundaries instead of normal droplet-like morphology (Fig.  6 d). This was accompanied by significant downregulation of the critical implantation marker COX2 (Fig.  6 e). Fig. 6 Effects of erastin-induced lipid peroxidation on endometrial receptivity and embryo implantation in mice. a Comparison of the average litter size between erastin-treated and vehicle-treated mice. Litter size data were collected from 10 vehicle-treated mice and 10 erastin-treated mice. The results are expressed as the mean ± SEM; **** P  < 0.0001. b Comparison of the average number of implantation sites on GD5 between erastin-treated and vehicle-treated mice. The results are expressed as the mean ± SEM; ** P  < 0.01. c Representative images of implantation sites (blue bands) on GD5 in erastin-treated and vehicle-treated mice. d HE staining of uterine sections at implantation sites on GD5 in erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. e Immunofluorescence staining of COX2 (green signal) co-stained with CK8 (red signal) and DAPI (blue signal) in uterine sections at implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice on GD5. f Immunofluorescence staining of the proliferation marker Ki67 (red signal) co-stained with E-cadherin (green signal) in uterine sections from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice on GD4. g RT-qPCR analysis of endometrial receptivity-related gene expression levels in erastin-treated ( n  = 5) and vehicle-treated ( n  = 5) mice on GD4. The results are expressed as the means ± SEM. ns, not significant; * P  < 0.05; ** P  < 0.01; *** P  < 0.001. h In situ hybridization staining of Ltf in uterine sections from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice on GD4. i Immunohistochemical staining of STAT3 and p-STAT3 in uterine sections from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice on GD4. Scale bars represent 200 μm and 100 μm Effects of erastin-induced lipid peroxidation on endometrial receptivity and embryo implantation in mice. a Comparison of the average litter size between erastin-treated and vehicle-treated mice. Litter size data were collected from 10 vehicle-treated mice and 10 erastin-treated mice. The results are expressed as the mean ± SEM; **** P  < 0.0001. b Comparison of the average number of implantation sites on GD5 between erastin-treated and vehicle-treated mice. The results are expressed as the mean ± SEM; ** P  < 0.01. c Representative images of implantation sites (blue bands) on GD5 in erastin-treated and vehicle-treated mice. d HE staining of uterine sections at implantation sites on GD5 in erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. e Immunofluorescence staining of COX2 (green signal) co-stained with CK8 (red signal) and DAPI (blue signal) in uterine sections at implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice on GD5. f Immunofluorescence staining of the proliferation marker Ki67 (red signal) co-stained with E-cadherin (green signal) in uterine sections from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice on GD4. g RT-qPCR analysis of endometrial receptivity-related gene expression levels in erastin-treated ( n  = 5) and vehicle-treated ( n  = 5) mice on GD4. The results are expressed as the means ± SEM. ns, not significant; * P  < 0.05; ** P  < 0.01; *** P  < 0.001. h In situ hybridization staining of Ltf in uterine sections from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice on GD4. i Immunohistochemical staining of STAT3 and p-STAT3 in uterine sections from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice on GD4. Scale bars represent 200 μm and 100 μm While uterine estrogen receptor (ER) and progesterone receptor (PR) expression patterns remained unaltered (Additional file 2: Fig. S5a, b) and FOXA2 immunostaining indicated preserved glandular function (Additional file 2: Fig. S5c), erastin treatment was associated with aberrant epithelial proliferation. Unlike controls, where Ki67 was restricted to stromal compartments, erastin-treated uteri showed epithelial Ki67 immunoreactivity (Fig.  6 f), indicating dysregulated cell cycle control without apoptotic changes (Additional file 2: Fig. S5d, e). On GD 4, qPCR analysis revealed dysregulated receptivity markers: significant upregulation of estrogen-responsive Ltf and downregulation of progesterone-responsive Areg and Ihh (Fig.  6 g). In situ hybridization confirmed elevated Ltf expression in uterine epithelium (Fig.  6 h), while immunohistochemistry showed increased STAT3 and phosphorylated STAT3 (p-STAT3, Tyr705) protein levels in endometrial epithelium (Fig.  6 i). Systemic assessment showed no ovarian impairment: serum E2 and P4 levels (Additional file 2: Fig. S6a, b), and ovarian expression of progesterone biosynthesis enzymes P450 side-chain cleavage enzyme (P450scc) and 3β-hydroxysteroid dehydrogenase (3β-HSD) (Additional file 2: Fig. S6c, d) were comparable between groups. These findings suggest that erastin-induced lipid peroxidation is associated with defective endometrial receptivity and embryo implantation, with no evidence of ovarian dysfunction. To evaluate maternal lipid peroxidation’s impact on decidualization, primary decidual zone (PDZ) formation was assessed at GD6 and GD8. Erastin-treated mice exhibited implantation failure or defective implantation sites with inter-individual variability (Fig.  7 a). Co-staining revealed significantly reduced expression of ZO-1 (a tight junction protein critical for cellular integrity) and FLK1 (the primary vascular endothelial growth factor receptor) at GD6 embryo implantation sites (Fig.  7 b). Concurrent Ki67/β-catenin co-staining (where β-catenin specifically marks the embryo implantation site) demonstrated diminished endometrial proliferation (Fig.  7 c). While vehicle controls showed sparse cleaved caspase-3 (CL-CASP3) expression, it was undetectable in erastin-treated mice (Fig.  7 d). In situ hybridization confirmed downregulation of decidualization markers Wnt4 , Bmp2 , and Hoxa10 at GD6 (Fig.  7 e–g). Fig. 7 Impaired primary decidual zone formation in erastin-treated mice. a Representative images of implantation sites (blue bands) on GD6 in erastin-treated and vehicle-treated mice. (b) Immunofluorescence co-staining for ZO-1 (red signal) and FLK1 (green signal) in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Nuclei were counterstained with DAPI (blue signal). c Immunofluorescence co-staining for Ki67 (red signal) and β-catenin (green signal, marking implantation sites) in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Nuclei were counterstained with DAPI (blue signal). d Immunofluorescence co-staining for cleaved caspase3 (CI-CASP3; green signal) and CK8 (red signal) in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Nuclei were counterstained with DAPI (blue signal). e In situ hybridization staining for Wnt4 in uterine sections on GD 6 implantation sites from erastin-treated and vehicle-treated mice. f In situ hybridization staining for Bmp2 in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. g In situ hybridization staining for Hoxa10 in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Scale bars represent 100 μm and 200 μm Impaired primary decidual zone formation in erastin-treated mice. a Representative images of implantation sites (blue bands) on GD6 in erastin-treated and vehicle-treated mice. (b) Immunofluorescence co-staining for ZO-1 (red signal) and FLK1 (green signal) in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Nuclei were counterstained with DAPI (blue signal). c Immunofluorescence co-staining for Ki67 (red signal) and β-catenin (green signal, marking implantation sites) in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Nuclei were counterstained with DAPI (blue signal). d Immunofluorescence co-staining for cleaved caspase3 (CI-CASP3; green signal) and CK8 (red signal) in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Nuclei were counterstained with DAPI (blue signal). e In situ hybridization staining for Wnt4 in uterine sections on GD 6 implantation sites from erastin-treated and vehicle-treated mice. f In situ hybridization staining for Bmp2 in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. g In situ hybridization staining for Hoxa10 in uterine sections on GD 6 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Scale bars represent 100 μm and 200 μm By GD8, erastin treatment induced embryonic morphological abnormalities, including disordered embryo spacing and reduced embryonic size (Fig.  8 a). Corresponding in situ hybridization showed suppressed expression of decidualization markers ( Bmp2 , Prl8a2 , Prl3c1 , Wnt4 , and Hoxa10 ) at implantation sites (Fig.  8 b–f). Collectively, these findings demonstrate that maternal lipid peroxidation severely impairs decidualization and disrupts early pregnancy progression. Fig. 8 Impaired decidualization in erastin-treated mice. a Representative images of implantation sites on GD8 in erastin-treated and vehicle-treated mice. b In situ hybridization staining for Bmp2 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. c In situ hybridization staining for Prl8a2 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. d In situ hybridization staining for Prl3c1 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. e In situ hybridization staining for Wnt4 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. f In situ hybridization staining for Hoxa10 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Scale bar represents 500 μm Impaired decidualization in erastin-treated mice. a Representative images of implantation sites on GD8 in erastin-treated and vehicle-treated mice. b In situ hybridization staining for Bmp2 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. c In situ hybridization staining for Prl8a2 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. d In situ hybridization staining for Prl3c1 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. e In situ hybridization staining for Wnt4 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. f In situ hybridization staining for Hoxa10 in uterine sections on GD 8 implantation sites from erastin-treated ( n  = 3) and vehicle-treated ( n  = 3) mice. Scale bar represents 500 μm

Infertility, defined as the failure to achieve a clinical pregnancy after ≥ 12 months of regular unprotected intercourse in reproductive-aged couples, affects approximately 15% of this demographic globally. Female factors contribute to ~ 60% of infertility etiologies [ 1 ]. Despite substantial advances in assisted reproductive technology (ART), infertility remains a critical public health challenge with profound physical, psychological, and socioeconomic consequences [ 2 ]. Embryo transfer—the pivotal assisted reproductive procedure for infertility treatment—confronts persistent post-transfer implantation failure, with 50%–60% of in vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI) cycles failing to achieve clinical pregnancy [ 3 , 4 ]. The establishment of early pregnancy is a multifactorial process that crucially depends on successful embryo implantation, subsequent development, and precisely regulated maternal–fetal crosstalk, thereby forming the foundation for a sustained gestation. Successful embryo implantation relies on three determinants: embryo quality, endometrial receptivity, and synchrony during the window of implantation (WOI). Cytokines, tissue factors, and hormones collectively modulate the endometrial microenvironment to influence pregnancy outcomes [ 5 , 6 ]. The etiology of recurrent implantation failure (RIF) involves multifactorial interactions across maternal, paternal, and embryonic domains, including immunological dysregulation [ 7 ], embryonic chromosomal abnormalities, uterine defects, endometritis, and endocrine disorders [ 8 – 11 ]. Within ART, while embryo quality from IVF/ICSI is necessary, endometrial receptivity is widely recognized as a predominant factor in implantation failure [ 12 , 13 ]. Lipid peroxidation, a central mechanism of oxidative cellular damage, is triggered by iron accumulation and excessive reactive oxygen species (ROS) [ 14 ]. It disrupts membrane integrity, fluidity, and permeability, causing redox imbalance. Excessive iron uptake via transferrin stimulates ROS via the Fenton reaction, promoting lipid peroxidation products; upon reaching a threshold, this induces ferroptosis [ 15 – 17 ]. Lipid-metabolizing enzymes (e.g., ACSL4, ALOXs) catalyze polyunsaturated fatty acid (PUFA) oxidation to lipid peroxides. These peroxides directly damage membranes or react with Fe 2+ to form •OH radicals, both directly triggering ferroptosis [ 15 ]. Resulting peroxides and metabolites drive a peroxidation cascade, potentially leading to tissue dysfunction. Specifically, malondialdehyde (MDA) originates from non-heme iron-mediated PUFA oxidation, while hydroxyeicosatetraenoic acids (HETEs) (e.g., 12-HETE, 15-HETE) amplify oxidative damage by regulating inflammatory factors [ 18 – 20 ]. Lipid peroxidation significantly impacts reproductive health in both sexes. In males, spermatozoa are highly susceptible to lipid peroxidation, resulting in structural and functional impairments and reduced fertility [ 21 , 22 ]. In females, elevated lipid peroxidation levels in ovarian tissue are associated with polycystic ovary syndrome (PCOS) [ 23 , 24 ] and may contribute to granulosa cell dysfunction and the development of primary ovarian insufficiency (POI) [ 25 ]. Uterine physiology is similarly affected, where glutathione peroxidase 4 (GPX4) deficiency in endometrial epithelial cells disrupts the lipid peroxidation balance and has been associated with implantation failure [ 26 ]. Furthermore, dysregulated lipid peroxidation has been implicated in various gestational complications, including preeclampsia, gestational diabetes mellitus, and intrauterine growth restriction [ 27 – 30 ]. Given the feasibility of assessing serum peroxidation and its established associations with human reproduction, this study further examined whether elevated maternal lipid peroxidation levels are associated with the occurrence of RIF and early pregnancy outcomes. Using a translational approach, we aimed to elucidate this relationship and its potential role in implantation and early pregnancy. Our goal was to evaluate the prognostic potential of lipid peroxidation and guide the development of future redox-modulating strategies for ART.

In the present study, we investigated the association between elevated maternal lipid peroxidation and both RIF and early pregnancy outcomes in a large cohort of patients undergoing IVF/ICSI. We found that elevated maternal lipid peroxidation is independently associated with RIF. Moreover, it is also independently linked to increased odds of implantation failure and early pregnancy loss after embryo transfer. This association was observed both in pre-transfer serum lipid peroxidation levels and in elevated intrauterine MDA levels on the transfer day. Supporting these clinical observations, murine models demonstrated heightened uterine susceptibility to lipid peroxidation damage compared to other organs, paralleling aspects observed in the clinical condition. Collectively, our findings suggest that heightened maternal lipid peroxidation is associated with increased risks of both implantation failure and pregnancy loss, potentially through effects on endometrial receptivity and decidualization. Lipid peroxidation, the oxidative degradation of PUFA-rich phospholipids, is initiated by the synergistic action of iron overload and ROS [ 41 ], driven by glutathione metabolism, iron homeostasis, and lipid regulation pathways [ 42 ]. Elevated levels of ROS, in the presence of catalytic Fe 2+ , trigger chain reactions that generate unstable lipid hydroperoxides (LOOH). Subsequently, LOOH decomposes via transition metal catalysis into reactive electrophiles, such as MDA and 4-HNE [ 43 , 44 ]. These reactive aldehydes can covalently modify macromolecules, including proteins and nucleic acids, thereby disrupting their structural and functional integrity [ 45 ]. Pathological lipid peroxidation and iron accumulation are implicated in the pathogenesis of various systemic disorders, including reproductive dysfunction [ 46 – 49 ]. Patients with RIF exhibited significantly higher serum MDA and non-heme iron levels than patients without RIF, suggesting potential dysregulation of lipid peroxidation. Further analysis of early pregnancy outcomes within the RIF cohort revealed that patients who achieved ongoing pregnancy had lower serum MDA levels and higher T-AOC than those who experienced pregnancy failure. Collectively, these observations suggest that lower lipid peroxidation levels and greater systemic antioxidant capacity may be associated with successful early pregnancy establishment in patients with RIF undergoing same-cycle transfer. These findings provide a rationale for further exploring redox-modulating approaches to support reproductive outcomes in this population. In-depth follow-up of early pregnancy outcomes in non-RIF patients revealed that the failed group exhibited significantly elevated serum lipid peroxidation compared to the successful group, with heightened peroxidation during the maternal implantation window associated with adverse same-cycle embryo transfer outcomes. These findings collectively indicate that elevated peri-implantation lipid peroxidation is associated with an increased risk of implantation failure and early pregnancy loss. Based on the above findings, we speculate that sustained elevation in lipid peroxidation across multiple embryo transfer cycles may constitute a potential contributing factor to the occurrence of RIF. Consequently, quantitative assessment of maternal lipid peroxidation before embryo transfer may have potential as a prognostic biomarker for cycle suitability and as a target for future investigations into redox-modulating interventions to optimize transfer success rates. Additionally, this study extends the exploration of the impact of local lipid peroxidation in the intrauterine microenvironment on early pregnancy after embryo transfer. The results suggest that elevated intrauterine lipid peroxidation levels are associated with increased odds of pregnancy failure in patients, and that maternal serum and intrauterine lipid peroxidation levels may be interrelated in affecting pregnancy outcomes after embryo transfer. Subgroup analysis revealed that the correlations between pre-transfer maternal lipid peroxidation levels and early pregnancy outcomes were significantly attenuated in patients with advanced maternal age (≥ 35 years), obesity (BMI ≥ 28 kg/m 2 ), PCOS, or endometriosis. Notably, lipid peroxidation showed no significant association with ongoing pregnancy in the obese and PCOS subgroups, indicating its limited utility as an independent prognostic marker in these specific populations. Mechanistically, these conditions are associated with distinct pathways that may elevate baseline oxidative stress: aging and obesity are linked to disrupted iron homeostasis and mitochondrial dysfunction, while PCOS is characterized by enhanced ovarian lipid peroxidation alongside suppressed antioxidant enzyme activity, which has been linked to granulosa cell dysfunction [ 23 , 24 , 50 – 52 ]. Concurrently, endometriotic lesions develop iron-saturated microenvironments from recurrent hemorrhage, driving lipid peroxidation-ferroptosis cascades [ 53 , 54 ]. We hypothesize that compensatory adaptations to persistent oxidative stress in these contexts may have altered the role of systemic lipid peroxidation, potentially rendering it more of an epiphenomenal biomarker than a primary pathogenic driver for implantation failure. It is plausible that other predominant pathological factors in these patients may exert a more direct influence on embryo implantation. This interpretation suggests that while systemic lipid peroxidation levels may not reliably predict reproductive prognosis in these specific subgroups, such stratified analysis provides crucial insights for the targeted management of heterogeneous infertility etiologies. Using a murine model of elevated lipid peroxidation during the implantation window, our study examined its association with multiple critical processes in early pregnancy. The experimental results suggest that elevated lipid peroxidation may be associated with impaired endometrial receptivity and decidualization, thereby compromising embryo implantation and early pregnancy progression. Estrogen and progesterone dynamics during early pregnancy critically regulate endometrial receptivity establishment [ 55 ] and are known to modulate redox homeostasis [ 56 ]. We hypothesize that high lipid peroxidation may impair the function of endometrial stromal and epithelial cells. This is supported by evidence that lipid peroxidation derivatives can induce protein carbonylation and disrupt functional protein activity [ 26 , 57 , 58 ]. We propose that these cellular and molecular disruptions may act synergistically, providing a potential mechanistic link to the observed implantation failure and early pregnancy loss. Our previous study identified uterine epithelial lipid peroxidation as a cause of infertility [ 26 ]. Clinically, however, systemic elevated lipid peroxidation (e.g., associated with inflammation) is more prevalent than loss-of-function GPX4 mutations in the endometrium. While small-sample studies reported elevated systemic MDA and non-heme iron in patients with RIF, the predictive value of pre-transfer serum lipid peroxidation for transfer outcomes remained unclear. To address this, the present large-scale study investigated the association between systemic lipid peroxidation and reproductive outcomes by analyzing serum and intrauterine fluid samples from an IVF/ICSI cohort. Using multiple assay parameters, we demonstrate that elevated maternal lipid peroxidation constitutes a significant and independent risk factor for implantation failure and early pregnancy loss, extending beyond RIF. Animal experiments revealed heightened uterine susceptibility to lipid peroxidation damage compared to other organs, more closely mimicking clinical conditions. This study has certain limitations. First, we did not directly compare maternal serum and intrauterine lipid peroxidation levels in the same patients, and any relationship between systemic and local peroxidation remains to be established. We speculate that differences or interactions between serum and uterine peroxidation could be relevant to endometrial receptivity and embryo implantation, but this requires further investigation. Second, the etiology of elevated lipid peroxidation in patients is complex, and intrinsic differences exist between animal models and the clinical population. Therefore, the relevance of our findings to both clinical data and animal experimental results requires careful interpretation. While our animal experiments provide preliminary evidence linking lipid peroxidation to impairments in endometrial receptivity and decidualization, these observations offer initial mechanistic insights rather than definitive proof. The specific regulatory pathways involved necessitate further verification and exploration in future studies. In addition, it would be valuable to further expand the number of patients in subgroup analyses to more reliably assess whether associations between maternal lipid peroxidation and pregnancy outcomes differ across patient subgroups. Finally, follow-up in the present study was limited to early pregnancy; longitudinal studies that include live-birth outcomes are still needed to achieve a more comprehensive assessment.

In conclusion, our findings demonstrate that elevated maternal lipid peroxidation is independently associated with the occurrence of RIF. Notably, in patients without RIF, it is specifically linked to implantation failure and early pregnancy loss. Experimental evidence indicates that the endometrium exhibits heightened sensitivity to fluctuations in maternal lipid peroxidation levels, which may adversely impact endometrial receptivity and decidualization, thereby potentially contributing to pregnancy loss. These findings underscore the critical role of lipid peroxidation homeostasis in embryo implantation and the maintenance of early pregnancy, and support further investigation into redox-modulating strategies in ART.

Additional file 1. Tables S1–S8. Table S1 Antibodies. Table S2 Probe primer sequences. Table S3 RT-qPCR primer sequences. Table S4 Comparison of clinical baseline characteristics between RIF and non-RIF groups before PSM. Table S5 Comparison of clinical baseline characteristics between the successful and failed pregnancy groups in patients without RIF before PSM. Table S6 Comparison of clinical baseline characteristics, lipid peroxidation, and antioxidant levels between the successful and failed pregnancy groups in patients without RIF. Table S7 Comparison of clinical baseline characteristics between the successful and failed pregnancy groups in patients from the cohort analyzed for intrauterine fluid. Table S8 Comparison of clinical baseline characteristics and MDA levels between the successful and failed pregnancy groups in patients from the cohort analyzed for intrauterine fluid after PSM. Additional file 2. Figures S1–S6. Fig. S1 Comparison of serum lipid peroxidation and antioxidant levels between the successful and failed pregnancy groups in patients with RIF. Fig. S2 Schematic diagram of erastin-induced lipid peroxidation and serum detection. Fig. S3 MDA detection in multiple organs. Fig. S4 Detection of uterine lipid peroxidation. Fig. S5 Lipid peroxidation does not affect steroid hormone receptors, glandular function, or apoptosis. Fig. S6 Lipid peroxidation does not affect ovarian function. Additional file 1. Tables S1–S8. Table S1 Antibodies. Table S2 Probe primer sequences. Table S3 RT-qPCR primer sequences. Table S4 Comparison of clinical baseline characteristics between RIF and non-RIF groups before PSM. Table S5 Comparison of clinical baseline characteristics between the successful and failed pregnancy groups in patients without RIF before PSM. Table S6 Comparison of clinical baseline characteristics, lipid peroxidation, and antioxidant levels between the successful and failed pregnancy groups in patients without RIF. Table S7 Comparison of clinical baseline characteristics between the successful and failed pregnancy groups in patients from the cohort analyzed for intrauterine fluid. Table S8 Comparison of clinical baseline characteristics and MDA levels between the successful and failed pregnancy groups in patients from the cohort analyzed for intrauterine fluid after PSM. Additional file 2. Figures S1–S6. Fig. S1 Comparison of serum lipid peroxidation and antioxidant levels between the successful and failed pregnancy groups in patients with RIF. Fig. S2 Schematic diagram of erastin-induced lipid peroxidation and serum detection. Fig. S3 MDA detection in multiple organs. Fig. S4 Detection of uterine lipid peroxidation. Fig. S5 Lipid peroxidation does not affect steroid hormone receptors, glandular function, or apoptosis. Fig. S6 Lipid peroxidation does not affect ovarian function.