The anti-Mullerian hormone as a predictor of early pregnancy loss in subfertile women

ABSTRACT

The aim of this study was to evaluate the predictive value of the anti-Mullerian hormone (AMH) level for early pregnancy loss and to compare the significance of AMH level to age as prognostic factors of pregnancy loss in subfertile women. The outcome of 848 subfertile patients confirmed with intrauterine pregnancies by ultrasound within 1 year of measuring serum AMH level were retrospectively analyzed. Among 848 patients, 206 women were diagnosed with early pregnancy loss. The mean age of the 848 patients was 35.66 ± 3.61 years (range: 26–46 years), and the mean AMH level was 2.95 ± 1.89 (range: 0.14–8.82 ng/mL). There were no significant differences in gravidity, parity, body mass index (BMI), and previous abortion history depending on early pregnancy loss. However, multivariable logistic regression analysis confirmed that the probability of early pregnancy loss is significantly affected by age (odd ratio, 1.079: 1.025–1.135, P = 0.004) and AMH (odd ratio, 0.885: 0.797–0.982, p = 0.022). According to this study, AMH level and age are both powerful predictors of early pregnancy loss. While chronological age is already well known as a factor related to early pregnancy loss, AMH was also considered when individualizing risk prediction for early pregnancy loss.

KEYWORDS:

• Anti-Mullerian hormone (AMH)
• age
• early pregnancy loss
• subfertility

Introduction

There is a well-known correlation between early pregnancy loss and maternal chronological aging. We also know that maternal chronological aging is associated with decreased ovarian reserve. Recently, a number of studies have been published that have validated the use of anti-Mullerian hormone (AMH) levels as an indicator of biological age (de Vet et al. 2002; Fanchin et al. 2003; Bungum et al. 2018)

The AMH is known to be the most precise predictor of ovarian reserve in women (de Vet et al. 2002; Fanchin et al. 2003). AMH is secreted by granulosa cells of early antral follicles (Vigier et al. 1984) and is associated with individual oocyte potential. Therefore, the more eggs a woman has, the greater the number of granulosa cells and the higher the AMH levels. AMH is barely detectable at birth and reaches its highest levels after puberty (Rajpert-De Meyts et al. 1999; Guibourdenche and Lucidarme 2003). Serum AMH levels decline with age, and the peak level occurs at 24.5 years of age. AMH levels are not affected by other hormonal variations, including the use of oral contraceptives (Kelsey et al. 2011). However, a recent study has revealed that oral contraceptives suppress FSH and LH secretion and, thus, inhibit follicular growth and cause low concentrations of AMH in long-term users. However, AMH increased by 53%, with values returning to normal within 2 months after discontinuation of oral contraceptives (Landersoe et al. 2020).

AMH levels were higher in the early follicular phase than in the late luteal phase, with a negative trend toward the luteal phase (Hadlow et al. 2013). With increasing chronological age, AMH levels decline and the aging ovary shows little variation in AMH level than does the younger ovary (Sowers et al. 2010). Correctly, the ‘aging ovary’ pattern was not related to chronological age but rather to relative ovarian age, as indicated by AMH (Sowers et al. 2010). During adulthood, although AMH continues to be expressed at basal and similar levels by both Sertoli and granulosa cells, its biological role is poorly understood (Anckaert et al. 2016). However, recent studies have found that AMH reflects the size of the pool of ovarian follicles (Durlinger et al. 2002; Gigli et al. 2005). During pregnancy, the overall trend of AMH levels declines in natural pregnancies. However, AMH levels either increased or decreased continuously from gestation week 4 to week 7 in pregnancy with assisted reproductive technology. Nonviable pregnancies showed unpredictable AMH patterns with sporadic changes, both increasing and decreasing in the same individual from gestation weeks 4–7 (Hamilton et al. 2016). Age-related reference values for AMH levels were obtained from pooled results, but these reference values have not yet been established for pregnancy (Anckaert et al. 2016).

Recently, many studies have examined the association between pregnancy complications and AMH levels (e.g., the relationship with pre-eclampsia or preterm births) (Kerkhof et al. 2010; Yarde et al. 2014; Stegmann et al. 2015). Yarde et al. demonstrated that women with a history of pre-eclampsia have significantly lower AMH levels than women with normotensive pregnancies (Yarde et al. 2014). Because pre-eclampsia is considered an indication of impaired vascular health, these results support the theory that compromised vascular health could act as a causative mechanism in early ovarian aging. Other studies have reported that biological age is more important than chronological age for female reproductive ability (Alviggi et al. 2009; Zhu et al. 2018). AMH is a novel marker of ovarian reserve and a good predictor of oocyte quantity (de Vet et al. 2002; Tremellen et al. 2005). Decreased AMH levels are associated with poor ovarian response to stimulation (Lie Fong et al. 2008). Recurrent miscarriages may be associated with diminished ovarian reserves. Recent prospective analyses indicated that larger prospective randomized controlled trials are warranted to better determine the predictive potential of ovarian reserve markers in recurrent miscarriages (Atasever et al. 2016; Pils et al. 2016; McCormack et al. 2019).

The aim of this study was to evaluate the predictive value of AMH levels for early pregnancy loss, and to compare the significance of AMH levels with age as a prognostic factor of pregnancy loss in subfertile women.

Results

A total of 848 patients were enrolled this study. The baseline characteristics are reported in Table 1. The mean age of the women was 35.66 ± 3.61 years and the mean BMI was 21.31 ± 2.8 kg/m². All patients were tested for AMH levels before pregnancy: mean 2.95 ± 1.89 ng/mL. Of the 848 patients, 206 (24.3%) had an early pregnancy loss within 12 weeks and 6 days of gestation. The number of patients who were pregnant naturally was 190 (22.41%), and the number of patients who conceived by IUI was 37 (4.37%). A total of 480 patients underwent IVF-ET (56.60%) and 141 (16.63%) patients received frozen embryo transfer (FET).

Table 1. Baseline characteristics of study population (N = 848)

Variable	Sub-category	Mean (min–max) or N	SD or %
Age (years)		35.66 (26–46)	3.61
BMI (kg/m^2)		21.31 (15.06–35.20)	2.8
Gravida	G1	543	64.03
	G2	194	22.88
	G3	79	9.32
	G4	21	2.48
	$\ge$G5	11	1.3
Para	P0	757	89.27
	P1	83	9.79
	P2	8	0.94
Spontaneous abortion	A0	608	71.7
	A1	153	18.04
	A2	59	6.96
	A3	18	2.12
	$\ge$A4	10	1.18
DM	(+)	6	0.71
HTN	(+)	8	0.95
Thyroid disease	(+)	107	12.59
AMH (ng/mL)		2.95 (0.14–8.82)	1.89
Pregnancy method	ART	658	77.59
	Natural pregnancy	190	22.41
1st trimester miscarriage	No	642	75.71
	Yes	206	24.29

SD: standard deviation; BMI: body mass index; DM: diabetes mellitus; HTN: hypertension; AMH: anti-Mullerian hormone; ART: assisted reproductive technology.

Table 2 presents patient demographic information including age, AMH, and other baseline characteristics of patients in both groups. There were no differences in gravidity, parity, BMI, and previous abortion histories. AMH levels were significantly lower in the early pregnancy loss group compared to the control group (2.52 ± 1.90 vs. 3.08 ± 1.87, P = 0.0002, respectively). There were also significant differences in age between the two groups (36.66 ± 4.07 vs. 35.34 ± 3.4, P < 0.0001, respectively). All patient’s serum FSH levels were evaluated on the second or third day of menstrual cycle. The mean FSH level was 8.90 mIU/mL in the control group and 9.72 mIU/mL in the early pregnancy loss group. The p-value of FSH level was 0.238 and was not statistically significant.

Table 2. Comparison of characteristics between control and early pregnancy loss groups

		Control (N = 642)	Early pregnancy loss (N = 206)
Variable	Sub-category	Mean(SD) or N(%)	Mean(SD) or N(%)	p-value
Age (years)		35.34(3.4)	36.66(4.07)	< 0.0001
BMI (kg/m^2)		21.25(2.72)	21.51(3.04)	0.2696
Gravida	G1	419(65.26)	124(60.19)
	G2	148(23.05)	46(22.33)
	G3	53(8.26)	26(12.62)	0.2942
	G4	15(2.34)	6(2.91)
	$\ge$G5	7(1.09)	4(1.94)
Para	P0	574(89.41)	183(88.83)
	P1	64(9.97)	19(9.22)	0.2266
	P2	4(0.62)	4(1.94)
Spontaneous abortion	A0	471(73.36)	137(66.50)
	A1	110(17.13)	43(20.87)
	A2	41(6.39)	18(8.74)	0.4383
	A3	13(2.02)	5(2.43)
	$\ge$A4	7(1.09)	3(1.46)
DM	(+)	2(0.31)	5(1.99)	0.0313
HTN	(+)	6(0.93)	2(1.00)	1
Thyroid disease	(+)	93(14.49)	13(6.31)	0.0022
AMH (ng/mL)		3.08(1.87)	2.52(1.90)	0.0002
FSH (mIU/mL)		8.90(7.80)	9.72(9.70)	0.238
Pregnancy method	ART	497(77.41)	161(78.15)	0.0013
	Natural pregnancy	145(22.59)	45(21.84)

SD: standard deviation; BMI: body mass index; DM: diabetes mellitus; HTN: hypertension; AMH: anti-Mullerian hormone; FSH: follicle stimulating hormone; ART: assisted reproductive technology.

Univariate and multivariate logistic regression analyses are described in Table 3. Univariate analysis found that age, AMH, chromosomal abnormality, and thyroid disease were factors significantly associated with early pregnancy loss (Table 3). When multivariate analysis was performed, aneuploidy was the only significant factor that correlated with early pregnancy loss, despite all variables included in this analysis. (P < 0.0001, data not shown). Therefore, multivariate logistic regression reanalysis was performed without aneuploidy factor.

Table 3. Univariate and multivariate logistic regression analysis of early pregnancy loss

	Univariate				Multivariate
	Odds ratio (95% CI)			p-value	Odds ratio (95% CI)			p-value
Age	1.107 (1.059–1.157)			<.0001	1.065 (1.011–1.122)			0.0186
BMI	1.032 (0.977–1.091)			0.2563	1.009 (0.949–1.073)			0.773
AMH	0.843 (0.77–0.922)			0.0002	0.884 (0.793–0.985)			0.0252
FSH	1.01 (0.993–1.028)			0.2612	1.002 (0.983–1.021)			0.8458
Gravida
G1 vs. others	0.805 (0.583–1.111)			0.1874	2.246 (0.696–7.245)			0.1757
Para
P0 vs. others	0.943 (0.571–1.556)			0.8171	0.515 (0.199–1.331)			0.1708
Spontaneous abortion
A0 vs. others	0.721 (0.514–1.011)			0.0578	0.344 0.114 1.034			0.0574
ART vs. NP
IVF-ET vs. NP	1.122 (0.758–1.661)			0.0532	0.857 (0.554–1.325)			0.1434
FET vs. NP	0.994 (0.594–1.662)			0.3274	0.771 (0.438–1.358)			0.4003
IUI vs. NP	0.391 (0.131–1.162)			0.0694	0.228 (0.051–1.014)			0.0726
Chromosome
Normal vs. abnormal	0.008 (0.003–0.017)			<.0001
HTN	0.938 (0.188–4.681)			0.9375	1.046 (0.196–5.57)			0.9579
DM	0.231 (0.051–1.042)			0.0566	1.223 (0.105–14.278)			0.8726
Thyroid disease	2.437 (1.332–4.456)			0.0038	2.218 (1.191–4.132)			0.0121

BMI: body mass index; DM: diabetes mellitus; HTN: hypertension; AMH: anti-Mullerian hormone; IUI: intrauterine insemination; IVF-ET: in vitro fertilization embryo transfer; FSH: follicle stimulating hormone; FET: frozen embryo transfer; NP: natural pregnancy; CI: confidence interval.

As a result, age, AMH, and underlying thyroid disease were found to be significantly associated with early pregnancy loss. (P-values were 0.0186, 0.0252, and 0.0121, respectively. Table 3)

Multivariate logistic regression analysis after adjusting for variables (BMI, gravidity, parity, abortion, hypertension, diabetes, and thyroid disease) to reveal the effects of age and AMH levels on early pregnancy loss are shown in Table 4.

Table 4. Logistic regression analysis of age and AMH in early pregnancy loss

	Odds ratio (95% CI)	p-value
Age ^(1)(years)	1.079 (1.025–1.135)	0.004
AMH ^(1)(ng/mL)	0.885 (0.797–0.982)	0.022

⁽¹⁾Adjusted variables: BMI, Gravida, Para, Abortion, DM, HTN, Thyroid disease.

CI: confidence interval; BMI: body mass index; DM: diabetes mellitus; HTN: hypertension; AMH: anti-Mullerian hormone.

We also analyzed the mean age and AMH levels between the aneuploidy group (N = 78) and euploidy group (N = 676). Of the 206 cases in the early pregnancy loss group, only 112 underwent chromosome analysis. Patients who did not confirm the chromosome analysis in the early pregnancy loss group were excluded from this analysis. Both age and AMH levels were significantly different between the groups. Age was higher in the abnormal group (37.88 ± 3.59 vs. 36.36 ± 3.47, P <.0001), whereas AMH was lower in the abnormal group (2.36 ± 1.76 vs. 3.06 ± 1.89, P = 0.002). (Table 5).

Table 5. Mean age and AMH of eupoild and aneuploid group

	euploid (N = 676)	aneuploid (N = 78)
	Mean ± S.D	Mean ± S.D	p-value
Age	36.36 ± 3.47	37.88 ± 3.59	<.0001
AMH	3.06 ± 1.89	2.36 ± 1.76	0.002

Additionally, we conducted logistic regression analysis to evaluate the relationship between maternal age and AMH with abnormal karyotyping (Table 6). Both age and AMH levels were significantly associated with chromosomal abnormalities in the product of conception.

Table 6. Logistic regression analysis of factor for chromosomal abnormality

	Odds ratio (95% CI)	p-value
Age	1.225 (1.143–1.314)	<.0001
AMH	0.802 (0.696–0.924)	<.0001

Discussion

Recently, women’s career advancement has become more important. Socially, as the age at which women get married is higher nowadays, the number of pregnancies at advanced maternal age is increased (KOSIS 2020). Some women have relatively good ovarian function despite their advanced age. On the contrary, others have diminished ovarian reserve despite their youth. The reproductive ability of women with decreased ovarian function results in an increase in assisted reproductive technology rates. Therefore, it is necessary to remind women with low AMH levels that they can predict a prognosis similar to that of an advanced-age pregnancy even though they are pregnant in their early 30s. Chronological age is an important predictor of pregnancy complications and miscarriage. However, recent studies have shown that biological age is more important in predicting complications of pregnancy than chronological age (Kerkhof et al. 2010; Yarde et al. 2014; Stegmann et al. 2015). This study was conducted to evaluate the risk of early pregnancy loss according to biological age represented by AMH in subfertile patients.

In natural conception, there was a prior study that revealed that the AMH levels had a reverse correlation with the risk of miscarriage. While AMH may not be an independent marker of fecundability, low AMH may be a marker of reduced reproductive potential due to its association with pregnancy loss (Lyttle Schumacher et al. 2018).

It has recently been demonstrated that women with recurrent miscarriages have lower AMH levels, which were even more pronounced in cases of idiopathic recurrent miscarriages (Pils et al. 2016; Atasever et al. 2016). Another retrospective study suggested that in women with idiopathic recurrent miscarriages and early pregnancy losses, AMH levels were positively correlated with the mean gestational age at miscarriage while patient age was not (Pils et al. 2017). Taratosconi et al. investigated whether serum AMH levels were independently associated with miscarriage rates after IVF-ET (Tarasconi et al. 2017). They observed that serum AMH levels were associated with miscarriage rates after IVF-ET and that such a relationship was not explained by usual age and oocyte yield fluctuations, as confirmed by multivariate regression analysis.

Prior studies have examined whether AMH is an indicator of oocyte quality. Since quality and quantity seem to go together during normal aging, it is conceivable that AMH might be a good predictor of embryo quality. One study showed that serum AMH concentrations on cycle Day 3 were correlated with the number of oocytes retrieved. However, AMH and embryo morphology were not correlated after conventional ovarian stimulation in IVF. Moreover, AMH and chromosomal competence of embryos were also not correlated. Hence, that study concluded that due to the lack of a consistent correlation with embryo morphology and embryo aneuploidy rates, there is no evidence for a direct relationship between oocyte quantity and embryo quality (Dewailly and Laven 2019). However, it is meaningful that AMH is related to an increase in aneuploidy in our study.

In this study, the AMH level within 1 year of pregnancy was used as an indicator of ovarian aging. In women aged 21 years and older, the average annual decline of AMH has been calculated to be 5.6% (Bentzen et al. 2013). Based on these findings, it is thought that AMH levels within a year of pregnancy in our study of pregnant women can be used as a criterion to reflect ovarian aging at the time of pregnancy. We also performed an analysis of FSH levels, which was used before AMH levels as an indicator of the reserve potential of the ovary. FSH is a hormone produced by the pituitary gland, an organ in the brain that enhances monthly follicular and oocyte growth and maturity. Levels of FSH are dependent upon a negative feedback system between the ovaries, hypothalamus, and pituitary gland. The results of this study showed that there were no significant differences in FSH values between the control and early pregnancy loss groups.

Most notable in this study was that although the risk of early pregnancy loss had been known to increase based on chronological age, biological age was also important. This study strengthens the concept that the prediction of pregnancy and miscarriage can be different according to the AMH values even at the same age. The strength of this study is that it included only subfertile patients with recent AMH levels (within 1 year before confirming pregnancy), and karyotype analysis of conception tissue was performed to evaluate the relevant cause of miscarriage. In fact, a significant correlation between AMH levels and miscarriage has been previously reported among patients undergoing IVF-ET (Tarasconi et al. 2017); however, no ensuing chromosome analysis data have been reported.

We initially analyzed all variables likely to be associated with early pregnancy loss by multivariate analysis. However, the results were absolutely influenced by chromosomal abnormalities. Chromosomal abnormalities seem to offset all associated factors, such as age and AMH. When analysis was performed in cases without chromosomal abnormality, age, AMH, and thyroid function were found to be related to miscarriage. In the logistic regression analysis results, we confirmed that maternal age and AMH levels are related to aneuploidy. Thus, among the various factors of early pregnancy miscarriage, we may conclude that factors such as age and AMH may be strongly associated with chromosome abnormality. Miscarriage can be caused by a variety of factors, but the most well-known cause of aneuploidy is maternal non-disjunction. Many hypotheses for the maternal age effect on non-disjunction are based on the degradation of the meiotic components in aged ovaries (Yoon et al. 1996). Another study has also shown that serum AMH level is a good predictor of ovarian response to controlled ovarian stimulation and fertilization as well as a good indicator of oocyte quality (Borges et al. 2017).

Previous studies have examined whether aneuploidy can be predicted with AMH levels. According to Plante et al., AMH levels did not differ between women with an aneuploid fetus and women with an euploid fetus (p = 0.46), and thus did not predict aneuploidy (Plante et al. 2010). Therefore, maternal AMH levels do not appear to be a marker of fetal aneuploidy in ongoing pregnancies. However, Shim et al. reported low maternal AMH levels as a predictor of fetal aneuploidy in early pregnancy loss (Shim et al. 2015). Results from previous data suggest that a higher biological age, indicated by lower AMH levels, was associated with very early types of pregnancy losses, although that study did not provide data on genetic abnormalities in their patient population (Pils et al. 2017). As in a previous report, our data suggest that the major reason for the increased miscarriage rate following low AMH, similar to the increase in age, might be the increase in chromosomal abnormality of the fetus.

It is well known that non-disjunction increases as the mother’s age increases, which in turn increases the risk of aneuploidy (Hassold et al. 1980; Kuliev et al. 2003; Munne et al. 2019). However, chromosomal non-disjunction in oocytes with low AMH levels has not been reported. Thus, further research is required to reveal the reason for the increased aneuploidy rate in patients with low AMH levels.

Another result of this study is that the presence or absence of thyroid disease, among underlying diseases, acts as an important factor in early pregnancy loss. We already know through previous studies that thyroid function plays an important role in embryonic health and development in early pregnancy (Krassas et al. 2015; Springer et al. 2017). We were able to reaffirm the results of this study.

To the best of our knowledge, this is the first study comparing ovarian reserve markers with age in women with early pregnancy losses. In this study, maternal AMH levels were lower in the early pregnancy loss group than in the control group, suggesting that AMH levels might be a valuable factor in predicting early pregnancy loss. While age is a powerful predictor of early pregnancy loss, low AMH should be considered as one of the risk factors for early pregnancy loss independent of age in subfertile patients. Larger prospective trials are warranted to establish the finding that lower AMH levels were associated with earlier gestational age at miscarriage and to further evaluate the value of AMH in the prediction of early pregnancy loss.

Materials and methods

Study design

A retrospective cohort study was designed to compare age and AMH as predictive values for early pregnancy loss in patients who had visited the CHA Gangnam Medical Center infertility clinic. The study was performed in patients who were confirmed with pregnancies by ultrasound and had a pre-pregnancy AMH result between January 2011 and December 2015 at Fertility Center of CHA Gangnam Medical Center. The study was conducted on patients with a history of subfertility, defined as those who tried to conceive for more than a year, but had not been successful and had visited the infertility clinic.

Patients who underwent hormone tests, including AMH results within 1 year from confirmation of pregnancy, were included in this study. Hormone levels were evaluated at menstrual cycle days 2 or 3. If positive results were confirmed in serum hCG, the ultrasound imagery was observed every 1–2 weeks to confirm normal pregnancy or early pregnancy loss. Pregnancy confirmation was defined as transvaginal ultrasound evidence of intrauterine gestational sac within the thickened deciduae. Early pregnancy loss was defined as a pregnancy lost after ultrasound documented an intrauterine pregnancy before 13 + 0 weeks of gestation, and dilation and curettage were performed. Of 1010 patients, 848 were included. Only patients who had full antenatal care at CHA Gangnam Medical Center were included in the study. Patients with polycystic ovary syndrome (PCOS) (N = 159) were excluded from the study based on AMH levels of more than 95% for each age group (Anckaert et al. 2016). In addition, patients with primary ovarian insufficiency (AMH levels<0.1, N = 3), severe male factor and uterine factor (severe adenomyosis or multiple myoma, huge myoma, and intrauterine adhesion) were also excluded. Patient characteristics, such as maternal age, gestational age, parity, reproductive history, method of conception, maternal body mass index (BMI), and AMH and FSH levels were reviewed. Pregnancy methods include natural pregnancy, intrauterine insemination (IUI), in vitro fertilization embryo transfer (IVF-ET), and frozen embryo transfer (FET). Each patient’s serum AMH level, serving as a marker of ovarian reserve, was measured at the infertility clinic.

AMH analysis

Serum AMH levels were quantified using the AMH Gen ll enzyme-linked immunosorbent assay (ELISA) (Beckman-Coulter, Inc., Brea, CA, USA). This assay is a standard monoclonal antibody sandwich enzyme immunoassay that is specific for AMH and does not exhibit any significant cross-reactivity with related molecules. All AMH patient levels were analyzed using this particular assay, which minimizes the error caused by using various assays. If the sample analysis was not completed within 48 h, the samples were frozen at −20°C.

Chromosome analysis

If a chromosomal analysis of the abortus was desired, cytogenetic analysis was performed first. However, in cytogenetic analysis of failed cases, the analysis was performed using multiplex ligation-dependent probe amplification (MLPA). In the study period, the products of conception samples were referred to the Genetics Laboratory of CHA Gangnam Medical Center. The samples were obtained after surgical evacuation or delivery due to spontaneous abortions at the CHA Gangnam Medical Center.

Statistical analysis

Statistical analyses were performed using IBM SPSS ver. 20 (IBM Corp., Armonk, NY, USA). Chi-square test and Student’s t-test were used for the calculation of significance. Descriptive data were expressed as the mean±standard deviation. Univariate and multivariate logistic regression analyses were performed to evaluate the association between parameters of ovarian reserve and early pregnancy loss. A P-value of <0.05 was considered statistically significant.

Ethics approval

This study was approved by the institutional review board of CHA Gangnam Medical Center, CHA University, Seoul, Korea. (GCI-19-23)

Authors’ contributions

Contributed to the study design, acquisition of data, analysis and interpretation of data, drafting of manuscript, and critical revision of the manuscript: SH; supervised the analysis, helped in the study coordination and the data collection: EC, EH; took part in the discussion and data analysis: SM, SK, MK, DC; participated in the design of the study and drafted the manuscript: SS, HP. All authors have read and approved the final manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.