High levels of anti-Müllerian hormone: what does this mean for IVF treatment?

Keywords::

• anti-müllerian hormone
• high ovarian response IVF
• ovarian hyperstimulation
• ovarian reserve controlled
• polycystic ovary syndrome

Is analysis of serum concentration of anti-Müllerian hormone (AMH) beginning to qualify as one of the crucial investigations in modern reproductive medicine, some 60 years after the French endocrinologist Alfred Jost suggested its existence? The number of studies on AMH reported on PubMed is increasing rapidly, and the clinical conditions for which the analysis is being considered vary substantially. Thus, analysis of AMH has been advocated for a number of diverse endocrine and clinical conditions: as a new and unifying test in the diagnostic procedure of polycystic ovaries and polycystic ovary syndrome (PCOS) [1], for predicting the time of menopause [2] and in the investigation of anovulation and amenorrhoea [3]. We will discuss its role in assisted reproduction techniques (ART), which is also the field in which the use of AMH analyses has become most widely spread, although some aspects have turned out to be controversial.

Although chronological age of the woman is an evident major determinant of the chance of delivering a live, healthy child after ART as well as in natural procreation, it has long been hypothesized that individual prognostic variation within age groups may depend on differences in ovarian ageing, or ovarian reserve. Ovarian reserve tests (ORTs) include various means to estimate the remaining follicle pool. Measuring basal serum levels of follicle-stimulating hormone (FSH) [4] was the earliest established test, being indirect in the sense that it relies on an intact hypothalamic–pituitary–ovarian feedback mechanism. Later, the combination of FSH levels with basal lutenizing hormone (LH) levels has proved to be more accurate in predicting ART outcome [5]. Still more reliable is direct antral follicle counting (AFC) by means of transvaginal ultrasonography [6]. Interestingly, reflecting an association between the number of antral follicles and the length of the follicular phase, information on the menstrual cycle length may be used as a crude estimate of ovarian reserve [7].

AMH is a dimeric glycoprotein, belonging to the TGF-β family. It is synthesised in growing preantral and small antral follicles and reflects the resting follicle pool. Serum AMH levels increase during puberty, peak at 24–25 years of age and then decline with age [8], being undetectable before menopause. AMH levels are generally analyzed by ELISAs.

In recent years, the predictive power of AMH assay in ART has been extensively studied. Although AMH levels are highly correlated with other ORTs, specifically with AFC, it has certain advantages. It is cycle independent [9,10], has a direct relationship with the resting follicle pool and is not dependent on the individual skill of the investigator, as is the case for scanning antral follicles. AMH levels, measured before the start of an in vitro fertilization (IVF) treatment, correlate strongly with follicular response to controlled ovarian hyperstimulation [11]. Expressed as the number of retrieved oocytes, or the number of retrieved oocytes divided by the total dose of FSH administered at controlled ovarian hyperstimulation, the ‘ovarian sensitivity index’, both excessive and poor ovarian responses may be predicted by AMH with great accuracy. It has been suggested that this index may be seen as a functional ORT [12,13].

The level of AMH is used to find the correct starting dose, thus minimizing the risks for suboptimal response in the case of a poor ovarian reserve and conversely, reducing the risk for ovarian hyperstimulation syndrome (OHSS) in women with a great ovarian reserve [14]. OHSS essentially does not occur in women with low AMH levels, whereas the risk for this syndrome increases dramatically with AMH levels above 4–5 ng/ml. Thus, with the analysis of a prestimulation AMH level, it is now possible to predict the patients who are at risk for this potentially life-threatening condition, and specific stimulation strategies may be chosen. These strategies involve very low starting doses of FSH (<100 IU per day) and a good deal of patience, acknowledging the slow initial growth of the follicles. The use of ‘coasting’, that is, withdrawing FSH in case of a threateningly high response should be used liberally as this strategy, applied for a limited number of days, does not impair ART results while it reduces risks for OHSS. The latest trend in stimulation strategies for high-AMH patients is to use a GnRH-antagonist protocol, with the explicit possibility to trigger the final oocyte maturation with a bolus dose of a GnRH agonist [15]. It is now evident that such a strategy, without the use of hCG, will effectively prevent OHSS. Usually, the embryos will all be cryopreserved in these cases, as the live birth rates will drop if an embryo was transferred directly. However, new rescue strategies involving a small supportive dose of hCG are promising [16].

Given the strong associations between AMH and ovarian response, and the association between the latter and ART outcome variables as pregnancy rates and live birth rates [17], it seems reasonable that AMH should also play a role in predicting such ultimate end points. Still, many studies on the subject have failed to find such relationships. To some extent, such results may be explained by a lack of statistical power because of small study populations and the reluctance to include women with the greatest ovarian reserve, those with polycystic ovaries and PCOS. The women with polycystic ovaries normally amount to a fifth of the ART population. Thus, when large ART populations are studied, covering the entire range of ovarian reserve including those with polycystic ovaries, measurements of ovarian reserve do show an age-independent association with pregnancy rates and live birth rates after IVF [18,19].

The distribution of AMH (and AFC) in large ART populations is log-normal, and the live birth rates are log-linearly associated with the AMH levels, resulting in some two- to three-times greater chance for a live birth in the women with levels above 2.94 ng/ml compared with those below 0.84 ng/ml [19]. It is important to acknowledge the log-linear relationship between AMH levels and ART outcome, meaning that even small changes in absolute figures in the low range of AMH (<1 ng/ml) may greatly influence outcome. In women with immeasurable serum AMH levels, extremely few IVF cycles result in a live birth (<5%) [19]. It should, however, be underlined that such pregnancies do occasionally occur, and thus there is no AMH level, at the current detection limit, for which the chance of success after ART is zero. This fact may result in complicated strategy discussions with couples belonging to this group.

Should women with polycystic ovaries be included in studies on ovarian reserve, or should they be considered a specific entity, not directly related to women with normal ovaries? Our studies clearly show that the polycystic ovary forms the high extreme in terms of ovarian reserve, seamlessly positioned after multifollicular ovaries and normal ovaries; and at the other extreme, the oligofollicular ovaries [5,7,18,19]. All the variables of ovarian reserve studied by our group (menstrual cycle length, FSH and LH levels, AFC and AMH levels) describe a (log-) Gaussian distribution, with polycystic ovaries at the right end. The LH:FSH ratio increases gradually to this end, as do menstrual cycle lengths, with anovulatory cycles becoming more and more prevalent in ovaries with the greatest number of antral follicles and the highest AMH levels. Thus, PCOS, as defined according to the Rotterdam criteria, becomes more prevalent with increasing AMH levels. When using the established criteria, PCOS would be diagnosed in about 25% of women with AFC ≥24 (corresponding to AMH levels above ~3 ng/ml). If the demands for AFC and AMH levels were to be increased to >38 and >5 ng/ml, respectively, that is, transferring the border for the diagnosis of polycystic ovaries further to the right as recently suggested [1], the proportion of cases with PCOS (in that subgroup) would increase as a result of the association between high AMH and both anovulation and hyperandrogenism. These findings underline the gradual changes in the clinical picture, in parallel with the morphologic and endocrine changes, in the spectrum of ovarian types and ovarian reserve [13].

Importantly, the findings suggest that women with polycystic ovaries, with a diagnosis based on AFC and AMH, are those with the best ovarian reserve, and they exhibit excellent ART results as a group, indeed with superior live birth rates compared with women with nonpolycystic ovaries, also after adjustment for age [18,19]. This said, one should also bear in mind that these women also constitute the group of patients who are at the greatest risk for OHSS. It may also be underlined that the patients in our ART population never exhibit BMI levels above 35, and mean BMI was between 23 and 24 kg/m² also in women with polycystic ovaries and PCOS. Thus, risks associated with morbid obesity often encountered in other studies on PCOS, were absent in this material.

An important finding in our recent study was that the association between AMH and live birth rates was still highly significant also after adjustment for both age and number of oocytes retrieved [19]. This finding, shown also for AFC [18], suggests that AMH (and AFC) captures not only a quantity aspect but also, to some extent, a quality aspect of the remaining oocytes. In support of this interpretation, are data showing an association between fecundability and AMH levels in natural procreation [20].

In conclusion, the analysis of AMH levels prior to ART may greatly affect the prognostic information given to a couple and influence stimulation strategies with specific possibilities to drastically reduce the incidence of OHSS. The highest AMH levels are found in women with polycystic ovaries, a group with the best treatment outcome, but also a group of patients who pose great challenges for stimulation strategies.

Financial & competing interests disclosure

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.