High serum Antimullerian hormone levels are associated with lower live birth rates in women with polycystic ovarian syndrome undergoing assisted reproductive technology

Anti-Müllerian hormone (AMH), or Müllerian inhibiting substance (MIS), is a member of the transforming growth factor beta (TGF-β) superfamily. It is an established indicator of ovarian reserve and predictor of ovarian response in assisted reproductive technologies (ART) [1, 2]. AMH plays a role in folliculogenesis by inhibiting primordial follicular recruitment and FSH-dependent growth and selection of antral follicles [3]. AMH is produced exclusively in the granulosa cells of ovarian preantral and small antral follicles. In women with polycystic ovarian syndrome (PCOS) serum AMH is elevated, around 2 to 4-fold higher than normal, which is thought to be implicated in the pathogenesis of PCOS.

Polycystic ovarian syndrome affects 5–8% of reproductive age women. Diagnostic hallmarks of PCOS include oligo or anovulation, hyperandrogenism, and antral follicular excess on ultrasound [4]. Not only is follicular excess contributing to elevated AMH, but granulosa cells in women with PCOS have also been shown to overproduce AMH [5]. AMH levels are associated with severity of PCOS and correlate with all three diagnostic hallmarks of the disease, making it potentially useful as a diagnostic criterion for PCOS [6‐12]. Based on the evidence for the association of AMH with PCOS pathophysiology, it has been investigated whether serum AMH may be related to assisted reproductive outcomes in women with PCOS.

There are many studies that have examined the link between serum AMH levels and IVF/ICSI outcomes such as implantation, clinical pregnancy, and live birth rate. AMH appears to be associated with IVF/ICSI outcomes but is overall a poor predictor of these outcomes [13‐15]. However, the results of various studies are inconsistent, and many of these did not include women with PCOS. In studies that have looked at these relationships in PCOS women, the results are conflicting. Although some studies have suggested a positive relationship [11, 16], some others have suggested an inverse relationship [17] between AMH concentration and pregnancy rates. Other studies have gone further to look at these associations based on PCOS phenotypes [18‐20], but no consensus has been reached. AMH is characteristically elevated in PCOS women and it is unknown whether it may predict ART outcomes in this population. Therefore, our goal was to determine whether serum AMH level is associated with ART cycle characteristics, pregnancy rates and live birth outcomes in women with PCOS.

The main baseline characteristics are listed in Table 1, and cycle characteristics are listed in Table 2. AMH levels decreased with increasing age of women in our cohort, with women in the low AMH group being significantly older compared to the average and high AMH groups. While the literature suggests an inverse relationship between BMI and AMH [22], there was little variation in BMI between the different AMH level groups in this study. In agreement with previous literature, a direct relationship was found between testosterone and AMH levels [11, 17] with a significant increase in the high AMH group compared to the average and low AMH groups. Total testosterone was 35.3 ± 10.2, 39.2 ± 12.8, and 58.0 ± 18.7 ng/dl in the <25th, 25 to 75th, and > 75th percentile, respectively. There was an inverse relationship between AMH levels and gonadotropin dose, which was significant in the <25th percentile group (3316 ± 1288 IU) compared to the 25 to 75th percentile (2587 ± 1071 IU) and > 75th percentile (2252 ± 842 IU) groups. E2 levels measured on the day of hCG administration varied among the groups with little evidence for a trend.

From the low to high AMH ranges, the average number of oocytes retrieved increased from 12.6 ± 6.3, 13.4 ± 6.2, to 14.2 ± 8.0, respectively. The number of fertilizations was highest in the high AMH range as well at 9.5 ± 5.6. Approximately two embryos on average were transferred for each group per procedure.

In this study, there was significant increase in live birth rate per embryo transfer (LBR) in PCOS women in the low AMH group (65.2%) compared to the average AMH (46.7%) and high AMH (43.5%) groups. The odds ratio (OR) for LBR in the low AMH group compared to the other groups (average and high) was 2.29 (p = 0.017) and remained significant after adjusting for age, BMI, day of transfer and number of embryos transferred (Table 3). Receiver-operating characteristic (ROC) curve analysis for AMH as a predictor of live birth revealed an area under the curve (AUC) of 0.543 (0.424–0.661, 95% confidence interval) (Fig. 1) indicating overall poor predictability for live birth. Although not statistically significant, the same inverse trend with AMH concentration was found for clinical pregnancy rate (CPR) (69.6, 53.2, 52.2%), implantation rate (IR) (47.1, 40.6, 35.3%), and multiple pregnancy rate (26.0, 21.7, 13.0%). Fertilization rate was highest in the low AMH range (70.4%), but there was not a uniform inverse relationship with this outcome.

This study is the first to describe an association between serum AMH levels and live birth outcomes in PCOS women. Live birth outcome is the ultimate clinically important measure in ART treatment and the results of this study indicate that PCOS women with higher serum AMH levels have poorer IVF outcomes compared to PCOS women with low/normal AMH levels.

In the present study, women with low serum AMH levels were older than those with average and high serum AMH levels. This relationship agrees with previous studies, as serum AMH levels begin declining around 25 years of age to undetectable levels at menopause [23]. Total gonadotropin dose was significantly higher in women in the low serum AMH group than those in the average and high serum AMH groups as well. FSH stimulates the growth and development of follicles in early folliculogenesis. The exogenous dose was dependent on intermittent daily measures of estradiol and ultrasound analysis of follicular growth. In the low serum AMH group, the increased gonadotropin dose compensated for lesser follicular growth by increasing the number of follicles that reach maturity. Although the number of oocytes retrieved increased slightly from low to high serum AMH groups, the success of this approach was reflected in the relatively similar number of oocytes retrieved across all women despite the broad range of serum AMH levels. This finding is supported by previous studies, showing that serum AMH levels reflect the ovarian response to stimulation [1, 24]. The fertilization rate was not significantly different between AMH groups, although the total number of fertilizations was also highest in the high serum AMH group, which could be attributed to greater number of oocytes retrieved.

In our cohort of women with PCOS undergoing assisted reproduction, LBR was significantly higher in the low serum AMH group (< 3.32 ng/ml, 65.2%) than those with average (3.32–8.27 ng/ml, 46.7%) or high serum AMH levels (> 8.27 ng/ml, 43.5%). Similar trends were noted in implantation, CPR, and multiple pregnancy rate, which each showing an inverse relationship with serum AMH levels, albeit non-significant. To our knowledge, this study is unique in that it demonstrates that higher serum AMH levels are associated with lower assisted reproduction LBR in women with PCOS. There have been several investigations into the association between AMH and ART outcomes in women with PCOS, but they did not include live birth rate as an outcome [11, 16, 17]. These studies reported conflicting results, some demonstrating a positive relationship between serum AMH levels and CPR [11, 16], while others having found an inverse relationship between serum AMH levels and CPR [17]. The reason for this discrepancy is unclear. Sample size may account for the variation in outcomes. Notably, in the study by Kaya et al. [16] there were significant age differences between the three AMH strata: Women in the high AMH (>75th percentile), average AMH (25th–75th percentile) and low AMH (<25th percentile) groups had mean age of 25.1, 29.2 and 34.1, respectively. This was in contrast to the study by Xi et al. [17] as well as the current study in which all three AMH groups were of more comparable age, which may explain some of the observed differences. Another possible explanation is the presence of multiple PCOS phenotypes and variable AMH concentration categorizations across studies. A 2015 meta-analysis of women with and without PCOS concluded that AMH was a weak predictor of assisted reproduction outcomes in non-PCOS women but not in women with PCOS [15]. There are a few recent studies that more directly examined the connections between PCOS phenotypes, AMH levels and associated ART outcomes. In one study, AMH was found to have a significant positive correlation with CPR and LBR only in one out of four designated PCOS phenotypes but not in the general PCOS population comprised of all 4 phenotypes [20]. Another study which focused on whether PCOS phenotypes were associated with cumulative LBR concluded that hyperandrogenic phenotypes confer significantly lower cumulative LBR [19]. Additional studies which focus on categorizing these phenotypes may help elucidate the exact associations between AMH concentrations and ART outcomes in women with PCOS.

There are a few possible explanations for the correlation of high AMH and poor IVF outcomes in PCOS women. AMH regulates folliculogenesis in the ovaries [25], and it is elevated as a hallmark of PCOS and correlates with the severity of PCOS manifestations [11]. The 2 to 4-fold increase in serum AMH levels reflects the follicular arrest and higher preantral and small antral follicle count. In addition, granulosa cells within each follicle have been shown to overproduce AMH up to 75-fold the normal amount [5, 26]. Elevated AMH may cause abnormal folliculogenesis through two different mechanisms [27]. AMH inhibits FSH-dependent selection of a dominant follicle, leading to decreased granulosa cell FSH sensitivity and potential follicular arrest [28]. In addition, prior studies showed a positive correlation between AMH levels and degree of hyperandrogenemia [7, 10, 11, 29]. AMH inhibition of granulosa cell FSH-induced aromatase expression can lead to an accumulation of intraovarian androgens [30, 31]. Moreover, androgens play a role in stimulating early (FSH independent) stages of follicular growth [32, 33] and may therefore contribute to increased AMH production in the ovary. These two mechanisms may contribute to PCOS phenotypes of hyperandrogenism and anovulation. In our study, there was a significant increase in total testosterone levels (58.0 ± 18.7 ng/dl) in the high serum AMH group when compared to total testosterone levels in the average (39.2 ± 12.8 ng/dl) and low (35.3 ± 10.2 ng/dl) serum AMH groups. Consistent with these observations, in the studies by Ramezanali et al. [20] and De Vos et al. [19] it was shown that hyperandrogenic phenotypes in PCOS women were associated with poorer CPR and LBR. Furthermore, there is previous research that suggests androgen excess is associated with poor oocyte quality, leading to lower fertilizations rates [34].

Not only have androgen levels been shown to directly correlate with elevated serum AMH, but studies have also shown LH to directly correlate as well. LH stimulates AMH production in granulosa cells in women with PCOS [5], which may contribute to AMH overexpression and AMH induced follicular arrest. Another proposed mechanism involves LH induced suppression of FSH, causing poor oocyte and embryo quality [17, 35], leading to low implantation and clinical pregnancy rates associated with high AMH levels.

PCOS has been shown to have a detrimental effect on endometrial homeostasis and receptivity [36]. Moreover, women with high serum AMH are at increased risk for developing ovarian hyperstimulation syndrome, which can abnormally affect endometrial receptivity, thereby affecting IR and LBR. There is additional evidence that suggests certain angiogenic factors, like VEGF and TGF-β, are dysregulated in PCOS [37], which could further affect the endometrium. We speculate that elevated levels of AMH seen in PCOS leads to aberrant folliculogenesis as well as possibly contributing to altered endometrial receptivity and abnormal placentation. There is evidence that placenta and endometrial cells express mRNA and protein for AMH and its specific type II receptor (AMHRII) [38, 39]. AMH is thought to exert negative effects on cellular viability by binding to its receptor [38, 40]. Interestingly, a recent study in PCOS women showed that higher AMH level was associated with decreased endometrial thickness in ovulation induction cycles [41]. Thus, endometrial interaction with elevated serum AMH levels could explain the associated low implantation and live birth rates.

This study is limited by its retrospective design as well as the relatively small sample size. In addition, it is important to mention that ovarian stimulation in our study’s patients was performed with caution to minimize the risk of ovarian hyperstimulation syndrome (OHSS), as reflected by the relatively similar number of oocytes retrieved in the various AMH patient groups as well as similar rates of OHSS. The patients in this study were treated during a period preceding the currently established protocols of GnRH antagonists for pituitary suppression and GnRH agonist trigger for those at risk of OHSS, and it is interesting to speculate whether a more aggressive approach to ovarian stimulation with freeze-all strategy would have led to a higher live birth rate in the high AMH group. This is suggested by a recent study showing that frozen embryo transfer is associated with higher LBR than fresh transfer in PCOS women [42]. Interestingly, De Vos et al. reported that live birth outcomes in PCOS women with hyperandrogenic phenotypes, as compared to other PCOS phenotypes, remain lower also as cumulative outcomes after taking into account both fresh and frozen embryo transfers [19]. Further larger studies looking specifically at LBR outcomes according to AMH levels and PCOS phenotypes in frozen embryo transfers using contemporary freeze-all cycle protocols should help to clarify the relationship.

In conclusion, this study found for the first time that PCOS women with low serum AMH levels (< 3.32 ng/ml) had significantly higher live birth rates than those with average or high serum AMH levels following ART. There were several other ART outcomes that correlated inversely with AMH concentrations. While the reason may not be clear, it may be associated with the positive link between AMH, testosterone and PCOS severity. The results of this study underscore the potential utility of AMH and other phenotypic features of PCOS rather than the basic diagnostic criteria of the disorder in treatment individualization and counseling in daily ART practice. PCOS is a multifaceted, heterogenous disease that is difficult to characterize and additional studies are necessary to fully understand how AMH and ART outcomes are related in various PCOS phenotypes. Such new knowledge will contribute to developing better strategies to improve the likelihood of a live birth in this special patient group.