Introduction

Polycystic ovary syndrome (PCOS) is the most common reproductive endocrine disorder in women, seen in 5–10% of women of reproductive age . It is characterised by a heterogeneous collection of signs and symptoms of variable severity. It is the commonest cause of anovulatory infertility and also associated with long-term health consequences. PCOS is defined by the presence of at least two out of three Rotterdam ESHRE–ASRM consensus criteria . These are oligo-anovulation and clinical and/or biochemical hyperandrogenism and the presence of polycystic ovarian morphology (PCOM) on ultrasound scan.

PCOM as diagnosed by ultrasound is common, with a prevalence of 20–30% in women of reproductive age . This has been defined as an ovary with at least 12 antral follicles between 2 and 9 mm in size . However, about half of the women with PCOM are asymptomatic and have no other criteria as detailed earlier to qualify for a diagnosis of PCOS.

Anti-Müllerian hormone (AMH) derives its name from its function in causing regression of the Müllerian duct, the female reproductive tract. In men, it is the first molecule to be synthesised and secreted by Sertoli cells during seminiferous tubule organisation. Sexual dimorphism occurs at 6–8 weeks of gestation in humans, and nothing remains of the Müllerian duct at 10 weeks although secretion of AMH persists until puberty in the male. Between the time of sexual dimorphism and puberty, the function of AMH remains unknown. AMH is a dimeric glycoprotein belonging to the transforming growth factor beta family produced by the granulosa cells of the ovarian follicles in women . It is expressed from as early as 36 weeks of gestation and continues throughout the reproductive lifespan to menopause. During this period, it is mainly expressed by the granulosa cells of the pre-antral and small antral follicles, which are involved in follicle-stimulating hormone (FSH)-dependent cyclic recruitment, with declining expression in the primordial, dominant and atretic follicles .

The AMH expressed by the antral follicles may be considered a good marker for the size of the antral follicle pool. Serum AMH levels show a strong association with the number of antral follicles . The density of pre-antral and small antral follicles in the polycystic ovary is six times that of the normal ovary . It is hence not surprising that serum AMH levels are higher in women with PCOS as compared to those without PCOS . Women with PCOM show significantly higher levels of serum AMH than those with normal ovaries but significantly lower levels than in women with PCOS. The women with isolated PCOM may be categorised as a separate entity midway between women with normal ovaries and those with PCOS according to AMH serum concentrations .

Normal physiology

While the exact function of AMH is yet to be completely clarified, there is evidence that it counteracts the actions of FSH on aromatase activity and in the development of an ovulatory follicle . A number of animal and human studies collectively demonstrate the inhibitory role of AMH in follicular growth. In murine granulosa cells, the FSH- and cyclic adenosine monophosphate (cAMP)-stimulated aromatase activity significantly reduced after treatment with AMH . The same study revealed reduced mRNA expression in cAMP-stimulated cells and decreased the number of luteinising hormone (LH) receptors in porcine granulosa cells with AMH treatment and that AMH inhibits factors that promote follicle progression and growth in human granulosa cells. Grynberg et al. (2012) demonstrated in humans that AMH expression can be differentially regulated by oestradiol depending on the oestradiol receptors suggesting that its decrease in the granulosa cells of growing follicles, which mainly express oestradiol receptor beta (ERβ), is due to the effect of oestradiol . Together, these effects of AMH would inhibit folliculogenesis or premature maturation in the normal ovary when the follicles remain at the small antral or antral stage. AMH production becomes almost undetectable once the size of the follicles is >10 mm , thereby allowing these follicles to become responsive to FSH and leading to follicle selection.

Normal physiology

While the exact function of AMH is yet to be completely clarified, there is evidence that it counteracts the actions of FSH on aromatase activity and in the development of an ovulatory follicle . A number of animal and human studies collectively demonstrate the inhibitory role of AMH in follicular growth. In murine granulosa cells, the FSH- and cyclic adenosine monophosphate (cAMP)-stimulated aromatase activity significantly reduced after treatment with AMH . The same study revealed reduced mRNA expression in cAMP-stimulated cells and decreased the number of luteinising hormone (LH) receptors in porcine granulosa cells with AMH treatment and that AMH inhibits factors that promote follicle progression and growth in human granulosa cells. Grynberg et al. (2012) demonstrated in humans that AMH expression can be differentially regulated by oestradiol depending on the oestradiol receptors suggesting that its decrease in the granulosa cells of growing follicles, which mainly express oestradiol receptor beta (ERβ), is due to the effect of oestradiol . Together, these effects of AMH would inhibit folliculogenesis or premature maturation in the normal ovary when the follicles remain at the small antral or antral stage. AMH production becomes almost undetectable once the size of the follicles is >10 mm , thereby allowing these follicles to become responsive to FSH and leading to follicle selection.

Pathophysiology in PCOS

FSH is key in follicular development, and serum FSH levels are observed to be significantly lower in women with PCOS than those with normal ovaries, although within normal limits . This in itself may be inadequate to explain the anovulatory process seen in PCOS. As FSH is responsible for the follicular development leading to ovulation, the oligo-anovulation, characteristic of PCOS, could be due to a dysfunction of FSH that may be quantitative, or qualitative or both ( Fig. 1 ).

A possible role for AMH in the anovulation associated with PCOS.

It is well accepted that the serum AMH levels reflect the number of antral follicles and hence are increased in women with PCOS and PCOM . A further study compared the ratio of serum AMH to the antral follicle count (AMH/AFC ratio) as a marker of AMH production per antral follicle in women with PCOS, PCOM and normal ovaries. This showed women with PCOS to have a significantly higher AMH/AFC ratio than those without PCOS. This indicated that women with PCOS had higher serum AMH levels because of not only a greater number of antral follicles but also greater AMH production per follicle . This is supported by in vitro studies, which have shown that individual granulosa cells from women with PCOS produce 75 times of AMH produced by the size-matched counterparts of women with normal ovaries . Interestingly, women with isolated PCOM and ovulatory cycles showed AMH/AFC ratios similar to women with normal ovaries. The AMH production per follicle has also been compared in the various phenotypes of PCOS. This showed that women with the anovulatory PCOS phenotypes (NIH classification types A, C and D) were shown to have significantly higher levels of serum AMH than the ovulatory PCOS phenotype with only PCOM and hyperandrogenism .

All of the above studies along with the reported property of AMH to counteract the actions of FSH in vitro imply that the high production of AMH by the polycystic ovary may have an important role in the pathophysiology of the syndrome . It could be hence hypothesised that the normal inhibitory function of AMH may be heightened in PCOS due to its increased intra-follicular concentration. Follicular unresponsiveness to FSH and failure of follicular development leading to anovulation may be affected by AMH-induced inhibition of aromatase, oestradiol and FSH and LH receptor acquisition.

Whether the higher AMH per follicle is a result of the influence of androgens, LH or insulin or whether it is an intrinsic property of the follicles of a polycystic ovary is not known. Numerous theories for the aetiology of anovulation associated with PCOS have been described, including an excess of small antral follicles, hyperandrogenaemia, hyperinsulinaemia and dysfunctional feedback mechanisms.

It has been hypothesised that androgens may drive excess AMH production in PCOS as, in serum, AMH is positively correlated with androgens . While the severity of hyperandrogenaemia seems to correlate well with the severity of the ovulatory disturbance, there is a paucity of evidence for a direct effect of androgens in the inhibition of ovulation. Hyperandrogenism is suspected to increase the AMH production by promoting an excess of small growing follicles and granulosa cell proliferation . However, the serum AMH levels were lower in women with hyperandrogenic normoovulatory PCOS than those with non-hyperandrogenic oligo-anovulatory PCOS . After 6 months of androgen suppression with dexamethasone, the AMH concentration remained unchanged . This could well imply that other factors may be involved in the production of excess AMH in association with PCOS other than androgens.

Hyperinsulinaemia is known to be present in anovulatory women more than ovulatory women and a direct correlation was found between serum AMH and insulin insensitivity . Hyperinsulinaemia increases androgen production, and the raised AMH concentrations may be secondary to an effect of insulin on the androgen levels. Evidence for a direct inhibitory effect of insulin on the ovulatory process is not available, and reduction in insulin levels with treatment was not followed by a decrease in the serum AMH levels .

While dysfunctional feedback mechanisms obviously play a part in the occurrence of anovulation, the changes in the feedback mechanisms are widely thought to be secondary to the primary pathophysiological changes in the ovary itself.

A plausible theory for the aetiology of PCOS is the developmental theory based on the principles of the Barker hypothesis. It is thought that exposure to high levels of androgens on the female foetus during intrauterine life will predispose to the development of PCOS following puberty. This idea originated from studies on pregnant Rhesus monkeys, injected with testosterone during pregnancy, and follow-up of the offspring who developed polycystic ovaries, oligomenorrhoea, high LH levels and insulin resistance. This theory is yet to be validated in humans.

Intrinsic polymorphisms or epigenetic changes in the AMH gene have been postulated but definitive evidence for the same is still awaited.

AMH as a diagnostic criterion for PCOS

AMH is produced by the pre-antral and small antral follicles, which when identified on ultrasound scan constitute the AFC. Due to this well-established strong correlation with the number of ovarian antral follicles , it has been suggested that serum AMH levels could be used interchangeably for the antral follicle count in the diagnosis of PCOS and integrated into the Rotterdam Criteria. Both variables, AMH and AFC, measure the same biological entity and hence if performed perfectly would yield similar values . Despite improvements due to technical advances, both these biomarkers still show weaknesses, which may affect their predictive ability to be used as a perfect diagnostic criterion.

The widespread availability of ultrasound to measure the AFC and immediate results from the procedure are its main advantages. However, intra-operator and inter-operator variability have been identified as the key limitation in its measurement . These are sufficient to cause changes to individual patient management and introduce bias when data from different centres are pooled for analysis and research . The introduction of the three-dimensional ultrasound with automated software and offline analysis of stored images has shown only a moderate reduction in variability with the added disadvantage of extra time and cost . The technical advances and variations in ultrasound machines available along with their widespread availability may have further compounded the problem of variability in results shown by different operators and clinics. With the introduction of new machines, it has now been possible to identify a greater number of antral follicles on scan. Using the existing definition of 12 follicles per ovary with an advanced ultrasound machine would thus lead to a significant increase in the incidence of PCOM and subsequently PCOS. Changing to a higher threshold of the AFC from the existing 12 follicles per ovary is probably necessary and has been suggested to overcome this problem .

An international standardisation detailing the technical requirements of the machine, frequencies of the ultrasound probes used and an external quality assurance is needed to ensure uniformity in the measurement of the AFC before a global diagnostic threshold is agreed.

By contrast, as a diagnostic criterion, serum AMH levels are advantageous over AFC as they are measured with a stable standardised assay reducing the inter-laboratory variability, especially with the newer automated assays . Serum AMH levels may also be considered useful in conditions where ultrasound scans for diagnosis may not be easily available. However, more than one AMH assay is available and values from these are significantly different from each other requiring an assay-specific interpretation .

Several studies have been done for assessing the accuracy of serum AMH as a diagnostic marker for PCOS and determining an optimal diagnostic threshold. There is still no consensus on the cut-off level of AMH for the diagnosis of PCOS. Various thresholds ranging from 2.8 ng/ml to 8.4 ng/ml have been proposed with varying sensitivities and specificities. A meta-analysis of these studies shows a cut-off level for serum AMH for diagnosis of PCOS at 4.7 ng/ml with a sensitivity of 79.4% and a specificity of 82.8% and an AUC of 0.87 .

However, it has still not been possible to incorporate the introduction of AMH into the definition of PCOS due to several issues. The studies done so far have shown heterogeneity with respect to the study quality characteristics, the study populations included, definitions of AFC and the AMH assay used . Most studies recruited women from fertility or gynaecology clinics rather than being population based, thereby leading to a potential problem in extrapolating these results to the general PCOS population. Some studies have used the definition of PCOM according to the 2003 Rotterdam criteria using a cut-off of 12 follicles without details of the ultrasound equipment used , whereas others have used higher cut-offs due to the use of a more sophisticated ultrasound equipment . There have been limitations in the comparison of these studies due to the different AMH assays used. In addition, the studies included diagnosed PCOS based on the Rotterdam criteria and hence the diagnostic thresholds suggested apply only to this subset of women. These values may not apply to women with PCOS based on the NIH/AES classification. Serum AMH levels correlate with the severity of hyperandrogenism and oligo-anovulation . Hence, it has been suggested that serum AMH levels may also be considered to replace these criteria for the diagnosis of PCOS . Women with PCOS show a variable phenotype. It has been suggested that AMH may not be suitable for the diagnosis of all phenotypes . Further studies to assess the accuracy of AMH for the diagnosis of different phenotypes and subgroups of PCOS are needed.

In conclusion, although serum AMH level is a very useful test for the identification of women with PCOS, the international standardisation of the AMH assay followed by large population-based studies leading to a consensus decision is needed before its routine incorporation into the diagnostic criteria for PCOS.