High AMH levels are associated with late miscarriage in women with PCOS undergoing frozen–thawed embryo transfer

Abstract

Background

Women with polycystic ovary syndrome (PCOS) have a high incidence of pregnancy and obstetric complications. Anti-Müllerian hormone (AMH) is closely related to reproductive outcomes. The potential relationship between AMH levels and the occurrence of late miscarriage in these patients remains unclear.

Methods

This study enrolled 1,831 patients diagnosed with PCOS who achieved clinical pregnancy following their first frozen-thawed embryo transfer cycles. The patients were categorized into miscarriage group and live birth group according to their pregnancy outcomes. The relationship between AMH levels and the occurrence of late miscarriage in this population was assessed before and after propensity score matching.

Results

PCOS patients with late miscarriage exhibited elevated AMH levels compared with those who had a live birth. According to univariate logistic regression analysis, each 1 ng/mL elevation in AMH level was related to a 6.3% increased risk of late miscarriage (odds ratio [OR], 1.063; 95% CI, 1.027–1.101; P = 0.001). After adjustment for confounding factors, the risk of late miscarriage increased by 5.6% (adjusted OR = 1.056, 95% CI: 1.012–1.101, adjusted P = 0.011). The odds of late miscarriage increased by 63.5% in the 50th percentile group (OR, 1.635; 95%CI, 1.129–2.369; P = 0.009), by 68.0% in the 75th percentile group (OR, 1.680; 95%CI, 1.141–2.474; P = 0.009) and by 93.8% in the 90th percentile group (OR, 1.938; 95%CI, 1.164–3.228; P = 0.011) when compared to the live birth group. After PSM, circulating AMH levels remained strongly linked to the occurrence of late miscarriage. The AUC of AMH for predicting late miscarriage was 0.614 (95% CI, 0.564–0.664; P = 0.000), with an optimal cutoff value of 10.395 ng/mL.

Conclusions

High pre-pregnancy circulating AMH levels (especially > 10.395ng/mL) is associated with a high risk of late miscarriage in patients with PCOS undergoing FET.

Trial registration

N/A.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-026-09348-y.

Introduction

Polycystic ovary syndrome (PCOS) is a highly prevalent reproductive and metabolic disorder that affects approximately 15% of women of reproductive age worldwide and is a major contributor to anovulatory infertility [1]. PCOS is a multifactorial condition, primarily characterized by ovulatory dysfunction, elevated anti-Müllerian hormone (AMH) levels, hyperandrogenism, and obesity, and is frequently accompanied by glucose and lipid metabolism disorders. While assisted reproductive technology (ART) serves as a well-established approach for managing PCOS-associated infertility, patients with PCOS tend to demonstrate a higher risk of adverse pregnancy and obstetric complications, including miscarriage, gestational hypertension, gestational diabetes mellitus, and preterm birth [2, 3].

AMH is a dimeric glycoprotein released by preantral and small antral follicles, and is widely recognized as a reliable marker of ovarian reserve. The 2023 international evidence-based guideline for PCOS have identified AMH as a diagnostic indicator for polycystic ovarian morphology in adult [4]. Recent studies suggest that AMH is closely related to female reproductive outcomes. Elevated AMH levels in preovulatory follicular fluid may forecast a greater likelihood of clinical pregnancy during in vitro fertilization (IVF) / intracytoplasmic sperm injection (ICSI) cycles [5]. Among women without PCOS, an AMH level < 1 ng/mL was independently associated with an elevated risk of miscarriage in fresh embryo transfer [6]. Additionally, a retrospective cohort study has demonstrated a significant correlation between elevated circulating AMH values and preterm delivery in PCOS patients [7]. Animal studies revealed that Thy1.2-AMH^Tg/0 mice, which overexpress AMH, exhibit markedly diminished reproductive capacity along with decreased litter sizes [8]. Notably, the administration of supraphysiological AMH doses in late gestation led to higher abortion rates and smaller litter sizes in mice [9], implying that excessive AMH concentrations during gestation may exert detrimental effects on fetal growth and development.

Late miscarriage is classified in China as complete pregnancy loss occurring between 12 and 28 gestational weeks. Although late miscarriage occurs less frequently than early miscarriage, it may result in greater psychological trauma to women and their families [10]. Additionally, complications following late miscarriage, such as intrauterine adhesions, cervical lacerations, hemorrhage, and infection, not only impair female reproductive potential and physical health but also contribute to the growing societal medical burden [11]. Prior researches have revealed that late miscarriage occurs more frequently in women with PCOS than in women without the condition, which might be attributed to abnormal endometrial function and poor placental quality [12, 13]. Evidence has also suggested that AMH could participate directly in trophoblast invasion and thereby influence placental function [14, 15]. However, the association between pre-pregnancy circulating AMH concentrations and late miscarriage among women with PCOS remains to be studied.

Accordingly, this study sought to elucidate the association between circulating AMH concentrations and late miscarriage among PCOS patients who underwent their first cycle of frozen-thawed embryo transfer (FET) and evaluate the risk of late miscarriage to identify the high-risk population.

Materials and methods

Participants

Ethical approval for this study was obtained from the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (approval No. 2017-KY-15), and written informed consent was acquired from each participant. Women diagnosed with PCOS who received their first FET cycle and achieved clinical pregnancy at the Reproductive Medicine Center of our hospital between January 2017 and June 2023 were obtained in this retrospective study. All data included in the analysis were obtained from the electronic medical records of our center.

The diagnosis of PCOS was established based on the 2003 Rotterdam criteria, which requires the presence of at least two of the three criteria: menstrual abnormalities (menstrual cycle > 35 days), hyperandrogenism (HA) and polycystic ovarian morphology (PCOM) [16]. PCOS can be categorized into four subtypes based on clinical features. Subtype A presents with HA + PCOM + menstrual abnormalities; subtype B shows HA + menstrual abnormalities; subtype C exhibits HA + PCOM and subtype D exhibits PCOM + menstrual abnormalities. We enrolled women aged 20–38 years who underwent their first FET cycle and subsequently attained clinical pregnancy. Participants meeting any of the following conditions were excluded: (1) cycles involving oocyte donation; (2) a history of ovarian surgery; (3) uterine abnormalities affecting embryo implantation, such as uterine malformation, submucosal uterine fibroids, intrauterine adhesion, endometritis and history of uterine surgery; (4) a history of cervical surgery or cervical insufficiency; (5) systemic diseases such as hypertension, autoimmune disease, malignant tumors, psychiatric disorder, thyroid dysfunction and other diseases affecting endocrine function; (6) smoking and drinking; (7) exposed to environmental pollutants in their living or working conditions; (8) repeated implantation failure and recurrent miscarriage; (9) incomplete data related to assisted reproductive treatment, pregnancies, delivery or AMH values. Ultimately, a total of 1,831 patients were analyzed in this study, among whom 128 had late miscarriage and 1,703 had a live birth.

Study protocols

Circulating hormone assessments were conducted at the specialized endocrinology laboratory of our reproductive medicine center within the 12 months before endometrial preparation, following the approaches described previously [17]. More specifically, circulating AMH concentrations were determined via the Roche Elecsys automated immunoassay system (RRID: AB_2895131). The IVF-ET protocols applied at our center have been documented in earlier publications, and patients were treated with natural cycles, hormone replacement treatment (HRT) cycles or GnRH agonist (GnRH-a) cycles for endometrial preparation and up to 2 embryos were transferred per patient [18, 19]. Clinical pregnancy was confirmed by the detection of at least one intrauterine gestational sac on ultrasound examination 35 days post embryo transfer. Late miscarriage was defined as spontaneous abortion between 12 and 28 weeks of gestation, while live birth referred to the delivery of viable neonates at a gestational age of 28 weeks or later. Trained nurses conducted telephone follow-ups to collect pregnancy and delivery-related outcomes for more than 12 months after embryo transfer.

Statistical analysis

Continuous variables were presented as mean ± standard deviations and comparisons between groups were conducted using the Student t-test. Categorical variables were summarized as frequencies and percentages and the chi-square test was used for comparison. Multivariable logistic regression analysis was utilized to investigate the relationship between serum AMH levels and the occurrence of late miscarriage. Adjusted variables included female age, BMI, serum baseline LH, AMH, total testosterone, TG, HDL, fasting insulin and the number of embryos transferred. Receiver operating characteristic (ROC) analysis was used to assess the predictive capacity for late miscarriage. The optimal cutoff value of AMH for predicting late miscarriage was established using the Youden index.

To mitigate confounding, propensity score matching (PSM) was performed to identify patients who achieved live birth and were comparable with those with late miscarriage. The propensity score was determined by using a multiple logistic regression model and BMI, baseline LH, total testosterone, TG and HDL were included as independent variables. The caliper value was 0.2. Patients with late miscarriage were matched to those who achieved live birth at a 1:8 ratio using the nearest neighbor matching algorithm, with matching conducted in a random order.

Statistical analyses were performed with IBM SPSS Statistics (version 26.0) and STATA Software/MP (Version 18.0). P<0.05 was deemed statistically significant.

Results

Baseline characteristics of PCOS patients with miscarriage and live birth

Baseline characteristics were retrospectively compared between late miscarriage and live birth groups. In our study, individuals experiencing late miscarriage exhibited higher BMI, elevated baseline serum LH, AMH, and total testosterone concentrations relative to those with live birth (P<0.05). Furthermore, the late miscarriage group also demonstrated higher triglyceride and lower HDL levels (P<0.05). Between-group differences in age, gravidity, parity, infertility duration, type of infertility, serum baseline FSH, PCOS phenotype, FT3, FT4, TSH, TC, LDL, FBG, fasting insulin, HOMA-IR, endometrial preparation, endometrial thickness on the day of embryo transfer, laboratory insemination, the number of embryos transferred and the stage of embryos transferred were not statistically significant (P>0.05) (Table 1).

Table 1.

Baseline characteristics of PCOS patients with late miscarriage and live birth before and after propensity score matching

Characteristics	Before PSM				After PSM
	Late Miscarriage (n = 128)	Live birth (n = 1703)	P	Late Miscarriage (n = 113)		Live birth (n = 759)	P
Age (y)	29.57 ± 3.79	29.28 ± 3.50	0.372	29.62 ± 3.72		29.31 ± 3.46	0.377
BMI (kg/m2)	25.11 ± 3.29	24.31 ± 3.61	0.016	25.25 ± 3.25		25.19 ± 3.57	0.855
Gravidity	0.78 ± 1.16	0.73 ± 0.92	0.138	0.76 ± 1.14		0.73 ± 0.91	0.755
Parity	0.23 ± 0.49	0.25 ± 0.46	0.764	0.24 ± 0.49		0.25 ± 0.47	0.833
Infertility duration (y)	4.45 ± 2.91	4.08 ± 2.67	0.138	4.55 ± 2.89		4.17 ± 2.66	0.166
Type of infertility			0.850				0.739
Primary infertility	63 (49.22%)	853 (50.09%)		57 (50.44%)		381(50.20%)	0.961
Secondary infertility	65 (50.78%)	850 (49.91%)		56 (49.56%)		378 (49.80%)
Baseline FSH (mIU/mL)	5.67 ± 1.47	5.74 ± 1.55	0.670	5.56 ± 1.40		5.65 ± 1.43	0.507
Baseline LH (mIU/mL)	9.32 ± 5.42	8.09 ± 5.02	0.012	10.05 ± 0.93		8.41 ± 0.31	0.318
AMH (ng/mL)	10.29 ± 5.10	8.83 ± 4.54	0.000	10.37 ± 5.41		8.84 ± 5.14	0.009
Total testosterone (ng/mL)	0.53 ± 0.26	0.43 ± 0.23	0.000	0.52 ± 0.26		0.49 ± 0.56	0.287
PCOS phenotype
A	59 (46.09%)	590 (34.64%)	0.060	54 (47.79%)		307 (40.45%)	0.492
B	2 (1.56%)	26 (1.50%)		2 (1.77%)		11 (1.45%))
C	1 (0.78%)	11 (0.63%)		1 (0.88%)		6 (0.79%)
D	65 (50.78%)	1076 (63.18%)		56 (49.56%)		435 (57.31%)
FT3(pmol/L)	5.18 ± 0.71	5.21 ± 0.68	0.681	5.22 ± 0.72		5.25 ± 0.64	0.718
FT4(pmol/L)	11.36 ± 1.98	11.41 ± 1.86	0.771	11.41 ± 2.01		11.33 ± 1.81	0.681
TSH (µIU/mL)	2.55 ± 1.26	2.48 ± 1.30	0.540	2.56 ± 1.28		2.54 ± 1.24	0.898
TG (mmol/L)	1.60 ± 1.45	1.34 ± 0.84	0.002	1.65 ± 1.51		1.51 ± 0.93	0.187
TC (mmol/L)	4.40 ± 0.84	4.35 ± 0.79	0.707	4.41 ± 0.86		4.31 ± 0.78	0.203
HDL (mmol/L)	1.32 ± 0.29	1.40 ± 0.34	0.012	1.29 ± 0.26		1.29 ± 0.28	0.869
LDL (mmol/L)	2.65 ± 0.76	2.61 ± 0.73	0.616	2.67 ± 0.78		2.64 ± 0.71	0.609
FBG (mmol/L)	4.94 ± 0.53	4.93 ± 0.45	0.820	4.97 ± 0.53		4.94 ± 0.46	0.589
Fasting insulin (µU/mL)	14.87 ± 9.37	13.30 ± 8.94	0.264	15.79 ± 9.76		13.58 ± 7.65	0.110
HOMA-IR	2.42 (1.96, 4.02)	2.47 (1.66, 3.44)	0.258	2.60 (2.02, 4.23)		2.63 (1.76, 3.69))	0.293
Endometrial preparation
Natural cycles	2 (1.56%)	59 (3.46%)	0.190	2 (1.77%)		28 (3.69%)	0.157
HRT cycles	96 (75.00%)	1336 (78.45%)		83 (73.45%)		594 (78.26%)
GnRH-a cycles	30 (23.44%)	308 (18.09%)		28 (24.78%)		137 (18.05%)
Endometrial thickness	9.56 ± 1.63	10.09 ± 1.79	0.053	9.46 ± 1.68		9.83 ± 1.78	0.160
Laboratory insemination			0.060				0.095
IVF	106 (82.81%)	1285 (75.46%)		92 (81.42%)		571 (75.23%)	0.151
ICSI	22 (17.19%)	418 (24.54%)		21 (18.58%)		188 (24.77%)
No. of embryos transferred	1.53 ± 0.50	1.47 ± 0.50	0.206	1.53 ± 0.501		1.45 ± 0.50	0.110
Stage of embryos transferred			0.322				0.993
Cleavage	43 (33.59%)	647 (38.00%)		40 (35.40%)		269 (35.44%)
Blastocyst	85 (66.41%)	1056 (62.01%)		73 (64.60%)		490 (64.56%)

Variables were presented as mean ± standard deviation or n (%)

HOMA-IR = fasting blood glucose (mmol/L) × fasting insulin (µU/mL) ÷ 22.5

BMI body mass index, FSH follicle stimulation hormone, LH luteinizing hormone, AMH anti-Müllerian hormone, FT3 free triiodothyronine, FT4 free thyroxine, TSH thyroid stimulating hormone, TG, triglyceride, TC total cholestero, HDL high-density lipoprotein, LDL low-density lipoprotein, FBG fasting blood glucose, HOMA-IR HOMA of insulin resistance, IVF in vitro fertilization, ICSI intracytoplasmic sperm injection

Association between circulating AMH levels and late miscarriage before PSM

Univariate logistic regression demonstrated that the risk of late miscarriage rose by 6.3% per 1 ng/mL increment in AMH concentrations (OR 1.063; 95% CI, 1.027–1.101; P = 0.001). Moreover, we performed multivariable logistic regression to evaluate the association between circulating AMH levels and late miscarriage. After adjusting for age, BMI, baseline LH, total testosterone, TG, HDL, fasting insulin and the number of embryos transferred, circulating AMH levels remained independently associated with late miscarriage, with an adjusted OR of 1.056 (95% CI, 1.012–1.101; P = 0.011) (Table 2). Serum total testosterone levels are also risk factors for late miscarriage (Table 2).

Table 2.

Key factors affecting the odds of late miscarriage

	OR (95% CI)	P	Adjusted OR (95% CI)	Adjusted P
Age (y)	1.023 (0.973, 1.077)	0.372	0.978 (0.883, 1.082)	0.661
BMI (kg/m2)	1.061 (1.011, 1.115)	0.017	1.032 (0.973, 1.095)	0.293
Baseline LH (mIU/mL)	1.047 (1.010, 1.086)	0.013	1.026 (0.988, 1.067)	0.181
AMH (ng/mL)	1.063 (1.027, 1.101)	0.001	1.056 (1.012, 1.101)	0.011
Testosterone (ng/mL)	3.923 (2.080, 7.401)	0.000	2.639 (1.248, 5.580)	0.024
TG (mmol/L)	1.256 (1.078, 1.463)	0.003	1.170 (0.980, 1.398)	0.083
HDL (mmol/L)	0.480 (0.271, 0.851)	0.012	0.504 (0.244, 1.041)	0.064
Fasting insulin (µU/mL)	1.023 (0.991, 1.056)	0.166	1.017 (0.977, 1.059)	0.420
No. of embryos transferred	1.261 (0.880, 1.808)	0.206	1.290 (0.873, 1.908)	0.202

Adjusted for age, BMI, baseline LH, total testosterone, TG, HDL, fasting insulin and the number of embryos transferred

BMI body mass index, LH luteinizing hormone, AMH anti-Müllerian hormone, TG triglyceride, HDL high-density lipoprotein, OR odds ratio

Furthermore, we stratified PCOS patients based on AMH percentile. We found that patients with AMH concentrations exceeding the 50th percentile (> 7.91 ng/mL) were more frequently identified in the miscarriage group compared to those in the live birth group (OR, 1.519; 95%CI, 1.054–2.189; P = 0.025). This trend was similarly observed in patients with AMH concentrations exceeding both the 75th (> 11.08 ng/mL) (OR, 1.580; 95%CI, 1.1077–2.318; P = 0.019) and 90th percentiles (> 15.14 ng/mL) (OR, 1.785; 95%CI, 1.078–2.955; P = 0.024). After adjusting for age, BMI, baseline LH, total testosterone, TG, HDL, fasting insulin, and the number of embryos transferred, the odds of late miscarriage increased by 63.5% in the 50th percentile group (OR, 1.635; 95%CI, 1.129–2.369; P = 0.009), by 68.0% in the 75th percentile group (OR, 1.680; 95%CI, 1.141–2.474; P = 0.009 ) and by 93.8.0% in the 90th percentile group (OR, 1.938; 95%CI, 1.164–3.228; P = 0.011) when compared to the live birth group (Table 3).

Table 3.

AMH breakdown of PCOS patients with late miscarriage and live birth before propensity score matching

AMH (ng/mL)	Late Miscarriage (n = 128)	Live birth (n = 1703)	P	OR (95% CI)	Adjusted P	aOR (95% CI)
Mean ± SD	10.29 ± 5.10	8.83 ± 4.54	0.001	1.063 (1.027, 1.101)	0.011	1.056 (1.012, 1.101)
> 25th (> 5.62)	105 (82.03%)	1268 (74.46%)	0.058	1.566 (0.985, 2.491)	0.285	1.314 (0.796, 2.168)
> 50th (> 7.91)	76 (59.38%)	835 (49.03%)	0.025	1.519 (1.054, 2.189)	0.009	1.635 (1.129, 2.369)
> 75th (> 11.08)	43 (33.59%)	413 (24.25%)	0.019	1.580 (1.077, 2.318)	0.009	1.680 (1.141, 2.474)
> 90th (> 15.14)	20 (15.62%)	161 (9.45%)	0.024	1.785 (1.078, 2.955)	0.011	1.938 (1.164, 3.228)

Adjusted for age, BMI, baseline LH, total testosterone, TG, HDL, fasting insulin and the number of embryos transferred

AMH anti-Müllerian hormone, OR odds ratio

Association between circulating AMH levels and late miscarriage after PSM

The baseline characteristics of patients after PSM were summarized in Table 1. The results indicated that after PSM, serum AMH levels remained significantly associated with the incidence of late miscarriage (OR, 1.070; 95%CI, 1.030–1.112; P = 0.000) (Table 4). We constructed a receiver operating characteristic (ROC) curve to evaluate the predictive value of AMH for late miscarriage. The area under the curve (AUC) was calculated to be 0.614 (95%CI, 0.565–0.664; P = 0.000), indicating that AMH was an independent predictor of late miscarriage in PCOS patients following FET (Fig. 1). The optimal AMH cutoff value was 10.395ng/mL.

Table 4.

AMH breakdown of PCOS patients with late miscarriage and live birth after propensity score matching

AMH (ng/mL)	Late Miscarriage n = 113	Live birth n = 759	P	OR (95% CI)
Mean ± SD	10.37 ± 5.41	8.84 ± 5.14	0.000	1.070 (1.030, 1.112)
> 25th (> 5.55)	91 (80.53%)	563 (74.18%)	0.045	1.653 (1.012, 2.699)
> 50th (> 7.94)	63 (55.75%)	373 (49.14%)	0.012	1.648 (1.117, 2.432)
> 75th (> 11.22)	36 (31.86%)	182 (23.98%)	0.007	1.748 (1.161, 2.631)
> 90th (> 15.54)	17 (15.04%)	70 (9.22%)	0.030	1.841(1.059, 3.201)

AMH anti-Müllerian hormone, OR odds ratio

Fig. 1.

Receiver operating characteristic curves for AMH in the prediction of late miscarriage after propensity score matching

Discussion

This retrospective cohort study investigated the association between high pre-pregnancy serum AMH levels and late miscarriage in patients with PCOS undergoing their first frozen-thawed embryo transfer. Among the PCOS population, individuals were classified into late miscarriage and live birth groups based on their pregnancy outcomes. The results of this study revealed a significant association between high AMH levels and an increased risk of late miscarriage. Furthermore, elevated levels of AMH were identified as a potential predictive factor for late miscarriage in PCOS patients.

It has been well documented that women with PCOS face a high risk of multiple pregnancy complications. Relative to the general population, PCOS is associated with a 2- to 4-fold elevation in the risk of pregnancy loss [20–22]. A large retrospective cohort study involving 21,820 pregnancies found that late miscarriage appeared to be more strongly associated with PCOS than early miscarriage [23]. Moreover, another study involving 15,210 pregnancies indicated that PCOS is independently associated with an elevated risk of late miscarriage following embryo transfer (OR 1.58, 95% CI 1.28–1.96) [24]. The studies referenced above demonstrate a close relationship between PCOS and the incidence of late miscarriage. Late miscarriage presents a deeply emotional and traumatic challenge for infertility patients seeking to take newborns home after undergoing IVF/ICSI treatment. While the clinical diagnosis of late miscarriage is relatively uncomplicated, the critical aspect lies in the early identification of its risk factors and the prompt implementation of appropriate interventions. Late miscarriage is associated with multiple causes, including uterine abnormalities, cervical insufficiency, infection or inflammation, and trauma [25, 26]. Nonetheless, up to the present, few specific biomarkers are available for identifying patients at risk. Early identification of risk factors for late miscarriage before ART treatment can aid in recognizing high-risk individuals, and the implementation of close targeted surveillance and timely interventions during pregnancy, when necessary, may provide significant benefits for these patients. In this retrospective cohort study, our analysis suggested that high AMH concentrations were correlated with the incidence of late miscarriage in PCOS patients following FET, particularly when AMH levels exceed 10.395ng/mL.

The clinical relevance of AMH continues to increase as more studies explore its predictive abilities regarding reproductive outcomes. Some previous researches have reported that elevated AMH concentrations may increase the risk of pregnancy loss. Through logistic regression models, Gleicher et al. found that extremely high AMH concentrations were significantly correlated with increased miscarriage rates in fresh embryo transfer cycles. When AMH levels increased from 5.5 ng/mL to 10 ng/mL, the miscarriage rate in the population under 36 years increased from 13.0% to 42.9%, while in the population over 43 years, the miscarriage rate increased from 61.1% to 81.8% [27]. However, they did not differentiate between early miscarriage and late miscarriage. A retrospective cohort study categorized 2246 patients into three groups according to their circulating AMH levels (low, medium, high), and they found that young women in the high-AMH group exhibited an elevated risk of early miscarriage relative to the medium-AMH group [28]. However, the association between elevated pre-pregnancy AMH concentrations and late miscarriage remains unclear, particularly in PCOS patients. In the present study, our analysis indicated that serum AMH concentrations were significantly elevated in PCOS patients who experienced late miscarriage compared with those who had a live birth.

To date, there are some indications that elevated AMH levels may be associated with placental dysfunction disorders. Stegmann et al. reported that stable or rising AMH levels in early pregnancy are positively correlated with an increased incidence of preterm delivery, and they suggested that AMH level screening in early pregnancy is necessary to identify preterm delivery incidence [29]. Among patients with PCOS, circulating AMH concentrations above the 75th percentile was related to an increased risk of preterm birth [7]. J.Y.Hsu et al. performed a retrospective cohort study and found that all women with PCOS and AMH levels above the 90th percentile had preterm deliveries [30]. Furthermore, our previous research indicated that high AMH levels are associated with gestational hypertension in PCOS patients [18]. Currently, animal experiments have demonstrated a direct association between elevated AMH levels and placental dysfunction. Ian S et al. found that Thy1.2-AMH^Tg/0 mice which express elevated levels of AMH produced no pups when mated with wild-type male mice, and spontaneous mid-gestational abortion and late resorption of fetuses were observed [8]. The potential biological mechanism may involve the participation of AMH in the process of trophoblast invasion. As a TGF-β superfamily cytokine, AMH can regulate the phosphorylation of SMAD 1/5/8 via AMH receptor II, share some biological mechanisms and functions with other cytokines in the TGF-β family, and may thus participate in trophoblast invasion [31, 32]. Therefore, we hypothesize that abnormally elevated levels of AMH in PCOS patients may lead to placental dysfunction, ultimately resulting in the occurrence of late miscarriage.

Furthermore, previous studies have suggested that AMH may be associated with cervical insufficiency, which is one of the significant causes of late miscarriage. Zhao et al. identified AMH as an independent predictor of cervical insufficiency among women with PCOS who underwent IVF-ET, with an optimal threshold value of 9.965 ng/mL for prediction [33]. A recent study further established a prediction model for cervical insufficiency among women undergoing IVF-ET and proposed AMH as an exploratory predictor [34]. Importantly, AMH has been shown to participate in extracellular matrix formation, tissue remodeling, and cellular proliferation [35, 36]. Based on the above background, we postulate that excessively high levels of AMH in PCOS may contribute to the development of cervical insufficiency and, thus, increase the risk of late miscarriage.

Our research found that increased AMH levels were associated with the development of late miscarriage in patients with PCOS, but ROC analysis showed that serum AMH level had limited predictive value for late miscarriage (AUC = 0.614). The possible reasons may include maternal metabolic conditions and hyperandrogenism caused by PCOS itself, which have affected the results. Accordingly, the comprehensive integration of serum AMH, metabolic risk indicators, and PCOS phenotypes in future assessments will be critical for more effectively stratifying the risk of late miscarriage among women with PCOS.

The strengths of this study include: (1) the first investigation exploring the link between AMH levels and late miscarriage among women with PCOS; (2) rigorous adjustments for confounding factors were achieved using PSM; (3) inter-cycle variation was reduced by restricting the analysis to the first embryo transfer cycle; (4) AMH measurements for participants were performed at our center prior to controlled ovarian stimulation, ensuring the comparability of outcomes among the participants.

It is important to acknowledge several limitations of this investigation: (1) the retrospective design excluded unmeasurable confounding factors, such as lifestyle and environmental variables; (2) single-center analysis, despite standardized clinical and laboratory workflows, may restrict the external validity of the findings; (3) although the overall sample size of this study is relatively large, the late miscarriage group consisted of only 128 women; (4) only PCOS patients undergoing their first FET cycles were analyzed in this study ; therefore, the results cannot be generalized to all patients with PCOS. Therefore, further prospective large-cohort studies are warranted for future investigation and discussion.

Conclusions

The results of this study suggest that AMH levels are high in patients with PCOS who experience late miscarriages after their first cycle of frozen-thawed embryo transfer. These individuals with high AMH levels prior to pregnancy face an increased risk of late miscarriage.

Abbreviations

PCOS

Polycystic ovarian syndrome

AMH

Anti-Mullerian hormone

IVF

In vitro fertilization

ICSI

Intracytoplasmic sperm injection

FET

Frozen-thawed embryo transfer

BMI

Body mass index

PSM

Propensity-score matching

Authors' contributions

Y.G. and Y.J. designed the study and revised the manuscript. Y.J. wrote the manuscript. Y. J., S.L., W. Z. and C. P. contributed to data collection and statistical analyses. F.W, Y.L. and Y.Z. contributed to paper revisiting.

Funding

This work was supported by the National Natural Science Foundation of China, Grant Number: 82471677; Key Research Project of Higher Education Institutions by the Henan Provincial Department of Education, Grant Number: 25A320017; Major Project of the Provincial and Ministry Collaborative Construction of the Henan Provincial Medical Science and Technology Research Program, Grant Number: SBGJ202501006.

Data availability

The data analyzed during this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study adhered to the Declaration of Helsinki regarding medical protocol and ethics and was approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (approval number 2017-KY-15). Informed consent was obtained from each participant.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.