Association of serum anti-Müllerian hormone levels with early miscarriage: a retrospective cohort and mendelian randomization study.

This study received ethical approval from the Ethics Committee of Peking University Third Hospital (Approval No.: IRB00006761- M2024908). Women who underwent IVF and achieved intrauterine clinical pregnancies in Peking University Third Hospital from January 2019 to December 2023 were enrolled. If a patient experienced multiple cycles during this period, only the first embryo transfer cycle was included. The exclusion criteria were as follows: (1) Endometrium thickness on the trigger day was less than 7 mm; (2) Presence of uterine abnormalities; (3) History of recurrent miscarriage; (4) Ovulatory dysfunction; (5) History of ovarian surgery; (6) Preimplantation genetic testing cycles; (7) Missing data on AMH. Our center has previously documented IVF procedures in detail [ 11 ]. Briefly, fertilization was achieved through either traditional IVF or intracytoplasmic sperm injection. Our specific methods for embryo vitrification and thawing have been outlined in prior research [ 12 , 13 ]. Up to two vitrified embryos were transferred either at the cleavage stage or blastocyst stage. To confirm biochemical pregnancy, we routinely measured serum β-human chorionic gonadotropin approximately 14 and 21 days after embryo transfer. Clinical pregnancy was confirmed via transvaginal ultrasound performed around 28 and 35 days after the transfer. The measurement of baseline serum AMH levels was previously described [ 14 ]. Serum AMH was measured by an ultrasensitive two-site ELISA assay (AnshLabs, Webster, TX, USA) within six months of initiating a started cycle. The coefficients of variation for inter-assay and intra-assay were less than 10% [ 15 ]. Early miscarriage was defined as the spontaneous termination of a clinical pregnancy before the completion of 12 weeks of gestation, with the fetus(es) being non-viable and not naturally absorbed or expelled from the uterus. Miscarriage or ongoing pregnancy was diagnosed by an experienced obstetrician-gynecologist through transvaginal ultrasound. The presence of an intrauterine pregnancy via ultrasound was confirmed before diagnosing a miscarriage. Restricted cubic spline (RCS) regression models were employed to characterize the nonlinear association between serum AMH concentrations and early miscarriage risk. A dual-threshold framework guided the analysis: (1) the statistical inflection point (odds ratio [OR] = 1) identified through RCS modeling [ 16 ], and (2) the established clinical threshold of 1.1 ng/mL derived from the previous research [ 4 ]. This hybrid approach ensured both data-driven detection of biological transitions and clinical interpretability aligned with reproductive medicine standards. Continuous variables were presented based on their distribution: means with standard deviations for normally distributed data and medians with 25th-75th percentiles for non-normally distributed data. Categorical variables were presented as frequencies with percentages. Student’s t-test was used to compare normally distributed continuous variables, while the Mann-Whitney U test was employed for non-normally distributed data. The chi-squared test was used to compare categorical variables. Logistic regression was utilized to compute OR with 95% confidence intervals (CI). Four models were performed by adjusting for different pre-specified confounding factors: (1) model 1 incorporated adjustment for female age alone as the established high correlation between AMH levels and female age; (2) model 2 adjusted for body mass index (BMI) and primary infertility; (3) model 3 combined adjustments for female age, BMI, and primary infertility; (4) model 4 comprehensively adjusted for female age, BMI, primary infertility, infertility etiology, endometrial thickness, cycle type, double-embryo transfer, blastocyst transfer status, and high-quality embryo transfer. Missing data were observed in the following variables: male age (0.002%), oocyte number (0.07%), infertile duration (0.05%), gravidity (0.003%), BMI (0.02%), endometrium thickness (38%), double embryo transfer (0.3%), and blastocyst transfer (2%). These missing values were imputed using fully conditional specification compatible with substantive models [ 17 ]. A subgroup analysis was conducted by analyzing fresh cycles or frozen cycles, women aged < 35y or women aged ≥ 35y. Sensitivity analyses were conducted using non-imputed data to assess result stability. Statistical significance was defined as a two-tailed p -value less than 0.05 in all analyses. The detailed procedure for MR analysis has been previously described [ 18 ]. Briefly, we extracted summary data from Genome-Wide Association Studies (GWAS) sourced from public datasets and existing research [ 19 – 22 ]. The comprehensive details and accessible links for each GWAS data were provided in Table S1 . The exposure trait was serum AMH levels and the measurement of AMH has been previously reported [ 19 ]. The outcome traits included spontaneous miscarriage and recurrent miscarriage, which were diagnosed based on the International Classification of Diseases codes ICD-O03 and ICD-N96, respectively. The MR analysis adhered to three fundamental principles: (1) the instrumental variable (IV) must be strongly associated with the exposure; (2) the IV should not be associated with any other potential confounders; and (3) the IV should not have a direct association with the outcome, instead influencing it solely through the exposure (Fig.  1 ). Based on these principles, we selected IVs from single nucleotide polymorphisms (SNPs) that met the following criteria: a genome-wide significance threshold of p  < 5 × 10 − 8 , absence of linkage disequilibrium (LD) validated through a clumping procedure (R 2  < 0.001, window size = 10,000 kb) to exclude strongly correlated variants and ensure SNP independence, exclusion of palindromic sequences, and an F-statistic value exceeding 10. All SNPs were scrutinized using the LDtrait database to confirm the absence of any associations between the IVs and potential confounders. Fig. 1 Flowchart of the study. A cohort study; B MR study. MR, mendelian randomization; SM, spontaneous miscarriage; RM, recurrent miscarriage; AMH, anti-Müllerian hormone; PGT, preimplantation genetic testing Flowchart of the study. A cohort study; B MR study. MR, mendelian randomization; SM, spontaneous miscarriage; RM, recurrent miscarriage; AMH, anti-Müllerian hormone; PGT, preimplantation genetic testing To ensure the robustness of our results, we employed several methods, including inverse variance weighted (IVW), MR-Egger, weighted median, simple mode, and weighted mode. The IVW result was considered the primary outcome due to its statistical power [ 23 ]. A p-value threshold of 0.05 was established for significance. Sensitivity analyses, including the ‘leave-one-out’ method, pleiotropy test, and horizontal pleiotropy test, were conducted to further validate the robustness of our findings. All analyses were conducted using R software (version 4.3.1).

Following the aforementioned criteria, a total of 19,154 participants who obtained intrauterine clinical pregnancy in Peking University Third Hospital from January 2019 to December 2023 were enrolled in this study (Fig.  1 ). RCS analysis identified an inflection point (AMH = 2.6ng/ml, Fig.  2 ). Thus, these women were categorized into three groups according to their AMH levels before IVF: AMH<1.1 ng/ml ( N  = 2,780), 1.1 < AMH ≤ 2.6 ng/ml ( N  = 6,666) and AMH ≥ 2.6ng/ml ( N  = 9,708). Baseline characteristics stratified by AMH levels are summarized in Table  1 . Women with diminished ovarian reserve (AMH < 1.1 ng/mL) were significantly older (34.63 ± 4.10 y) compared to those with intermediate (1.1–2.6 ng/mL; 33.21 ± 3.96 y) and normal AMH levels (≥ 2.6 ng/mL; 32.02 ± 3.69 y). A parallel age gradient was observed in male partners, with the oldest spouses in the low AMH group (35.71 ± 5.26 y), followed by intermediate (34.43 ± 5.00 y) and normal AMH cohorts (33.36 ± 4.75 y). BMI exhibited incremental decreases across AMH strata, ranging from 23.01 ± 3.59 kg/m² in the lowest AMH group, 22.62 ± 3.68 kg/m² in the intermediate AMH group, to 22.21 ± 3.53 kg/m² in the highest. The prevalence of primary infertility increased progressively with AMH levels (55.1% vs. 60.5% vs. 63.5%). Duration of infertility was longer in low AMH groups (2-5y) compared to the intermediate and normal AMH group (2–6y). Tubal factor infertility plateaued in higher AMH groups (35.1% vs. 40.6% vs. 40.6%), whereas endometriosis prevalence declined with increasing AMH (11.1% vs. 8.2% vs. 7.0%). Male factor infertility decreased markedly across AMH strata (46.9% vs. 40.5% vs. 33.0%). Number of oocytes was more in higher AMH groups (10–17) compared to the intermediate and low AMH group (7–12, 5–9, respectively). Frozen cycles increased progressively with AMH levels (37.7% vs. 48.9% vs. 67.8%). Fertilization methods shifted significantly across groups, with conventional IVF utilization declining (71.0% vs. 62.1% vs. 58.0%) and ICSI increasing proportionally (27.6% vs. 34.8% vs. 37.0%). Endometrial thickness showed minimal variation (9–11 mm vs. 9.3–12 mm vs. 9–11 mm) with borderline significance. Blastocyst transfer rates increased with AMH levels (26.3% vs. 30.4% vs. 35.4%), while high-quality embryo rates slightly declined (85.3% vs. 84.2% vs. 83.1%). Post-transfer biochemical markers exhibited AMH-dependent elevations, with median hCG levels rising from 850.66 mIU/mL (day 14) and 14,003 mIU/mL (day 21) in the lowest AMH group, 896.57 mIU/mL and 14,613 mIU/mL in the intermediate AMH group, to 982.32 mIU/mL and 15,542 mIU/mL, respectively, in the highest. Fig. 2 Relationship between AMH and early miscarriage in RCS analysis. A Unadjusted. B Adjusted for age. C Adjusted for BMI and primary infertility. D Adjusted for age, BMI, and primary infertility; E Adjusted for female age, BMI, primary infertility, cause of infertility, endometrial thickness, cycle type, double ET, blastocyst transfer, and high-quality ET. BMI, body mass index. RCS, restricted cubic spline; ET, embryo transfer Relationship between AMH and early miscarriage in RCS analysis. A Unadjusted. B Adjusted for age. C Adjusted for BMI and primary infertility. D Adjusted for age, BMI, and primary infertility; E Adjusted for female age, BMI, primary infertility, cause of infertility, endometrial thickness, cycle type, double ET, blastocyst transfer, and high-quality ET. BMI, body mass index. RCS, restricted cubic spline; ET, embryo transfer Table 1 Characteristics of women categorized based on their AMH level <1.1 ng/ml ( N  = 2,780) 1.1–2.6 ng/ml ( N  = 6,666) ≥ 2.6 ng/ml ( N  = 9,708) p Female age, mean (SD), y 34.63 (4.10) 33.21 (3.96) 32.02 (3.69) < 0.001 Male age, mean (SD), y 35.71 (5.26) 34.43 (5.00) 33.36 (4.75) < 0.001 BMI, mean (SD), kg/m2 23.01 (3.59) 22.62 (3.68) 22.21 (3.53) < 0.001 Primary infertility 1,531 (55.1) 4,032 (60.5) 6,167 (63.5) < 0.001 Infertile duration, median (IQR), y 3 (2, 6) 3 (2, 5) 3 (2, 5) 0.029 Cause of infertility  Tube factor 976 (35.1) 2,705 (40.6) 3,940 (40.6) < 0.001  Endometriosis 308 (11.1) 546 (8.2) 681 (7.0) < 0.001  Male factor 1,304 (46.9) 2,701 (40.5) 3,205 (33.0) < 0.001  Other factor 720 (25.9) 1,820 (27.3) 3,183 (32.8) < 0.001  Oocyte number, median (IQR) 7 (5, 9) 10 (7, 12) 13 (10, 17) < 0.001 Cycle type  Fresh 1,731 (62.3) 3,403 (51.1) 3,122 (32.2) < 0.001  Frozen 1,049 (37.7) 3,263 (48.9) 6,586 (67.8) Fertilization method < 0.001  IVF 1,973 (71.0) 4,138 (62.1) 5,631 (58.0)  ICSI 768 (27.6) 2,321 (34.8) 3,588 (37.0)  Unclear 39 (1.4) 207 (3.1) 489 (5.0) Endometrium thickness, mean (SD), mm 10 (9, 11) 10 (9.3, 12) 10 (9, 11) 0.002 Double ET 1,706 (61.6) 4,315 (64.9) 6,003 (61.9) 0.001 Blastocyst transfer 730 (26.3) 2,025 (30.4) 3,436 (35.4) < 0.001 High-quality ET 2,370 (85.3) 5,613 (84.2) 8,063 (83.1) 0.011 HCG level after ET  14 days, median (IQR), mIU/ml 850.66 (489.64, 1,597) 896.57 (510.44, 1,801) 982.32 (554.03, 1,975.75) < 0.001  21 days, median (IQR), mIU/ml 14,003 (7700, 22,695) 14,613.5 (8,306.25, 23,194.75) 15,542 (8,984, 24,600.5) < 0.001 Unless indicated otherwise, values are presented as numbers (percentages) ET Embryo transfer, AMH anti-Müllerian hormone, SD standard deviation, BMI body mass index, IVF in vitro fertilization, ICSI intracytoplasmic sperm injection, ET embryo transfer, HCG human chorionic gonadotrophin Characteristics of women categorized based on their AMH level Unless indicated otherwise, values are presented as numbers (percentages) ET Embryo transfer, AMH anti-Müllerian hormone, SD standard deviation, BMI body mass index, IVF in vitro fertilization, ICSI intracytoplasmic sperm injection, ET embryo transfer, HCG human chorionic gonadotrophin RCS analysis revealed a significant non-linear association between AMH levels and early miscarriage risk, demonstrating a progressive decline in aOR with increasing AMH concentrations across the full cohort ( p  < 0.001) (Fig.  2 ). Subgroup analyses of the fresh cycle and frozen cycle were consistent with this trend (Fig S1 −2). Among women aged<35y, the relationship between ORs of early miscarriage and AMH levels exhibited a U-shaped pattern, with ORs decreasing initially and then increasing (Fig S3). In contrast, among women aged ≥ 35y, ORs of early miscarriage decreased significantly with increasing AMH levels (Fig S4). The same method was applied without imputing missing values, yielding consistent results (Fig S5-8). Reduced AMH levels demonstrated a dose-dependent association with early miscarriage risk. Women with AMH < 1.1 ng/mL exhibited the highest incidence (17.7% vs. 10.9% reference; aOR = 1.26 [1.11–1.44], p  < 0.001), followed by those with AMH 1.1–2.6 ng/mL (13.8% vs. 10.9%; aOR = 1.13 [1.02–1.25], p  = 0.018) (Table  2 ). Subgroup analyses across clinical strata, including fresh embryo transfer cycles, frozen embryo transfer cycles, and age-stratified cohorts (< 35 and ≥ 35 y), demonstrated consistent directional associations. Although statistical significance levels fluctuated following adjustment for subgroup-specific confounders, all aORs remained above unity (aOR > 1), indicating persistent risk elevation across all evaluated clinical scenarios. These results remained consistent in the analysis without data imputation (Table S2). Table 2 Logistic regression of early miscarriage and AMH Group <1.1 ng/ml 1.1–2.6 ng/ml ≥ 2.6 ng/ml Ratio OR Ratio OR Ratio OR Total  Raw 17.70% 1.76 (1.57–1.98) 13.80% 1.31 (1.2–1.44) 10.90% Ref  Model 1 - 1.38 (1.22–1.57) - 1.17 (1.06–1.29) - Ref  Model 2 - 1.71 (1.52–1.92) - 1.3 (1.18–1.43) - Ref  Model 3 - 1.36 (1.2–1.54) - 1.16 (1.06–1.28) - Ref  Model 4 - 1.26 (1.11–1.44) - 1.13 (1.02–1.25) - Ref Fresh  Raw 18.40% 1.92 (1.63–2.27) 14.00% 1.39 (1.2–1.62) 10.50% Ref  Model 1 - 1.36 (1.14–1.63) - 1.19 (1.02–1.38) - Ref  Model 2 - 1.89 (1.6–2.24) - 1.38 (1.19–1.6) - Ref  Model 3 - 1.36 (1.13–1.63) - 1.19 (1.02–1.38) - Ref  Model 4 - 1.28 (1.06–1.54) - 1.19 (1.02–1.39) - Ref Frozen  Raw 16.50% 1.59 (1.33–1.91) 13.60% 1.27 (1.12–1.44) 11.00% Ref  Model 1 - 1.32 (1.09–1.59) - 1.15 (1.01–1.31) - Ref  Model 2 - 1.54 (1.28–1.85) - 1.25 (1.1–1.42) - Ref  Model 3 - 1.29 (1.07–1.56) - 1.14 (1–1.3) - Ref  Model 4 - 1.14 (0.94–1.38) - 1.07 (0.94–1.22) - Ref Age < 35y  Raw 12.60% 1.38 (1.16–1.66) 10.60% 1.14 (1–1.29) 9.40% Ref  Model 1 - 1.34 (1.12–1.61) - 1.12 (0.99–1.27) - Ref  Model 2 - 1.36 (1.14–1.63) - 1.13 (0.99–1.28) - Ref  Model 3 - 1.32 (1.1–1.58) - 1.11 (0.98–1.26) - Ref  Model 4 - 1.23 (1.02–1.49) - 1.09 (0.96–1.24) - Ref Age ≥ 35y  Raw 22.40% 1.6 (1.35–1.89) 19.30% 1.33 (1.14–1.55) 15.30% Ref  Model 1 - 1.35 (1.13–1.61) - 1.23 (1.05–1.43) - Ref  Model 2 - 1.57 (1.32–1.85) - 1.32 (1.13–1.53) - Ref  Model 3 - 1.33 (1.12–1.59) - 1.22 (1.05–1.42) - Ref  Model 4 - 1.21 (1.01–1.46) - 1.15 (0.99–1.35) - Ref OR odds ratio, CI confidence interval, AMH anti-Müllerian hormone, BMI body mass index, ET embryo transfer Model 1 adjusted for female age Model 2 adjusted for BMI and cause of infertility Model 3 adjusted for female age, BMI, and cause of infertility Model 4 adjusted for female age, BMI, primary infertility, cause of infertility, endometrial thickness, cycle type, double ET, blastocyst transfer, and high-quality ET Logistic regression of early miscarriage and AMH OR odds ratio, CI confidence interval, AMH anti-Müllerian hormone, BMI body mass index, ET embryo transfer Model 1 adjusted for female age Model 2 adjusted for BMI and cause of infertility Model 3 adjusted for female age, BMI, and cause of infertility Model 4 adjusted for female age, BMI, primary infertility, cause of infertility, endometrial thickness, cycle type, double ET, blastocyst transfer, and high-quality ET Following the steps outlined to achieve the three core assumptions of MR, we identified six SNPs associated with AMH levels as IVs (Table S3). Our analysis did not detect a causal relationship between AMH levels and spontaneous miscarriage (OR = 1.11 [0.90–1.36] and OR = 1.00 [1.00–1.00], respectively) or recurrent miscarriage (OR = 0.94 [0.86–1.04]) (Fig.  3 , Table S4). Neither the MR-Egger method nor the IVW method detected heterogeneity (Table S5). The robustness of our findings was further confirmed by visual inspection of the forest plot, funnel plot, and results from the ‘Leave-one-out’ method (Fig S10-13). Additionally, no evidence of pleiotropy was detected through the MR Egger intercept and the MR-PRESSO global test (Table S6). Fig. 3 Casual relationships between AMH and miscarriage in MR analysis. AMH, anti-Müllerian hormone; MR, mendelian randomization; SM, spontaneous miscarriage (O15_ABORT_SPONTAN); SM2, spontaneous miscarriage (ukb-d-O03); RM, recurrent miscarriage Casual relationships between AMH and miscarriage in MR analysis. AMH, anti-Müllerian hormone; MR, mendelian randomization; SM, spontaneous miscarriage (O15_ABORT_SPONTAN); SM2, spontaneous miscarriage (ukb-d-O03); RM, recurrent miscarriage

Early miscarriage, defined as the loss of pregnancy at any gestational age less than 12 weeks, occurs in 11.9% of pregnancies [ 1 ]. Pregnant women who experience early miscarriage often bear great mental stress and psychological burdens in the early stages, making them more susceptible to anxiety, depression, and other psychiatric disorders [ 2 ]. Compared to naturally conceived pregnancies, women who conceived by in vitro fertilization (IVF) cycles are more concerned with the progression of their pregnancies and the likelihood of achieving a live birth. Therefore, exploring potential serum biomarkers to predict the risk of early miscarriage can help clinicians and patients in identifying this risk early, providing scientific guidance for embryo transfer populations, and enhancing care and management for those who achieve a pregnancy. Anti-Müllerian hormone (AMH), as one of the indicators for ovarian reserve [ 3 ], has been suggested to be associated with the risk of early miscarriage. A meta-analysis of 13 retrospective studies enrolling 12,042 women found that lower serum AMH levels were associated with an increased risk of miscarriage (adjusted odds ratio [aOR] = 1.35 [1.10–1.66]) [ 4 ]. However, it is noteworthy that differences in AMH detection techniques and varying thresholds for defining high and low serum AMH levels among studies limit the reliability and applicability of these findings. Evidence regarding the association between AMH and early miscarriage remains controversial. In recent three-year studies, some research supports that lower pre-pregnancy AMH levels were related to the increased risk of miscarriage, possibly due to decreased oocyte quality [ 5 – 7 ]. Conversely, other studies indicate no association between pre-pregnancy serum AMH levels and miscarriage [ 8 – 10 ]. Currently, there is no definitive conclusion regarding the relationship between serum AMH levels and miscarriage. Therefore, our study seeks to validate the relationship between serum AMH levels and early miscarriage within a large sample of IVF populations. Additionally, we aim to identify the causal effects of serum AMH levels on miscarriage by employing two-sample Mendelian randomization (MR).

Our stud y revealed that low serum AMH levels in IVF might be associated with higher rates of early miscarriage. However, no causal relationship was established between AMH levels and miscarriage in women who conceived naturally. Recent studies in IVF cycles have shown that the predictive power of AMH levels for miscarriage differs between patients with polycystic ovary syndrome (PCOS) and those without the diagnose [ 7 ]. Specifically, in non-PCOS patients, an AMH level < 1 ng/ml is associated with a significantly increased risk of miscarriage, which aligns with our findings. However, no such association was observed in patients with PCOS [ 7 ]. A possible explanation for this disparity in PCOS patients is that increased levels of androgens, luteinizing hormone, and other undefined factors lead to an increase in the number of small follicles and subsequent elevated AMH secretion, which may not accurately reflect ovarian reserve or oocyte quality [ 24 – 26 ]. Previous studies in IVF populations have not adequately addressed this issue by excluding this relevant subset of patients [ 9 , 27 ]. In our study, we further excluded patients with ovulation disorders and a history of ovarian surgery to ensure a homogeneous population. This study further employed MR analysis to address the limitations associated with the observational nature of the research, thereby minimizing confounding factors and reverse causation. The results of the MR analysis indicated that AMH levels were not related to the incidence of spontaneous or recurrent miscarriages. Moreover, in the results of the cohort study, the aOR for early miscarriage were very close to 1 after adjusting for age, indicating a minimal association between AMH levels and early miscarriage in women who underwent IVF. However, it should be noted that the GWAS populations for the exposure SNPs did not exclude individuals with ovulatory dysfunction or a history of ovarian surgery, and the GWAS populations for the outcome SNPs comprised solely of naturally conceived pregnancies. Therefore, women with low AMH levels who conceive naturally do not need to worry excessively about early miscarriage. However, in the IVF population, low AMH levels warrant caution regarding the risk of early miscarriage. To our knowledge, this study is the first to investigate the relationship between AMH levels and early miscarriages using a combination of cohort studies and MR. We conducted RCS analysis to illustrate how the rate of early miscarriages changes as AMH levels increase continuously. Furthermore, we performed several regression models and a series of sensitivity analyses to ensure the robustness of our findings. The sample size was substantial, and our data were relatively complete, with minimal missing information. The GWAS summary data for AMH utilized in this study are the most recent and largest available. We collected GWAS data for outcome traits from various databases to ensure consistency in our conclusions, and conducted heterogeneity tests and sensitivity analyses to ensure the reliability of our results. However, it is crucial to acknowledge the limitations of this study. In the cohort analysis, the primary limitation stems from its retrospective nature, which may have introduced residual and unmeasured confounding factors. Despite adjusting for available covariates in the multivariable analysis and conducting several sensitivity analyses, we could not fully account for all potential confounding effects and exclude the residual confounding effects. Furthermore, the retrospective design limited our ability to collect comprehensive and relevant information. Additionally, excluding patients who had undergone ovarian surgery may have impacted the applicability of the results to some extent. However, given the clear impact of ovarian surgery on ovarian function, excluding these patients helped us to better elucidate the relationship between AMH levels and early miscarriage. In the MR analysis, firstly, our research exclusively included individuals of European ancestry. This may result in disease incidence and AMH levels that are specific to this group, limiting the generalizability of our conclusions to other populations, particularly given that our cohort consists of Chinese IVF patients. Nonetheless, we incorporated as much outcome data as possible to enhance the reliability of our findings and their applicability to various European populations. Currently, GWAS data for the early miscarriage traits is not available, prompting us to select spontaneous miscarriage and recurrent miscarriage as outcome traits. Due to the constraints of GWAS summary-level data, we were unable to conduct subgroup analyses to investigate potential stratified effects of factors such as age, PCOS, and ovarian surgery history. While we have made efforts to ensure that the SNPs used in our study were not associated with confounding factors in the published research, there may be potential confounding factors that affect the relationship between AMH and miscarriage, warranting further exploration.

Low serum AMH levels may be associated with increased risks of early miscarriage in women who conceive through IVF. However, there is no causal relationship between low AMH levels and miscarriage in naturally conceived women. This conclusion necessitates further research for validation.

Supplementary Material 1 Supplementary Material 1