Ovarian disorders - technical article

Topics covered:

• Essentials
• Introduction
• The hypothalamic–pituitary–ovarian axis
• The menstrual cycle
• Disorders of ovulation - includes detailed section on polycystic ovarian syndrome
• Other causes of hyperandrogenism in women
• Further reading

Ovarian development and folliculogenesis

Ovarian development is essentially complete by about 6 months of fetal life, and at this time the ovaries contain some 6–7 million germ cells. By the time of birth, the number of germ cells has fallen to 1–2 million and the remaining germ cells have entered the first meiotic division to form oocytes. Each oocyte is surrounded by a single layer of flattened, somatic pregranulosa cells that forms the primordial follicle. It is the primordial follicles that constitute the resting pool, which must provide sufficient oocytes to last a normal reproductive lifespan.

Ovarian organogenesis and follicle formation

The fetal ovary is formed from three embryonic cell lineages: the coelomic epithelium, the mesenchyme of the mesonephros (primitive kidney), and primordial germ cells (which arise within the extraembryonic tissue of the yolk sac). Around embryonic day 35, the coelomic epithelium thickens over the mesial aspect of the mesonephros, forming the gonadal ridge. The underlying mesenchymal cells of the mesonephros also divide, and the gonadal ridge protrudes into the coelomic cavity as the gonadal anlagen or primordium. Concurrently, the primordial germ cells begin to migrate from the yolk sac using an amoeboid action. Before and during migration they divide by mitosis and an estimated maximum of 1700 enter the primordium. Once populated by primordial germ cells, the primordium becomes the indifferent gonad. This initial development of the ovary is identical to that of the testis until morphological changes occur at around day 39 of embryonic life that make the male and female gonads distinguishable. Evidence from the mouse indicates that once they have reached the gonad, primordial germ cells differentiate and lose their migratory ability; they are then known as oogonia.

Differentiation of the ovary occurs at embryonic day 40 to 42, a few days after that of the testis. Oogonia dramatically increase in number by mitosis, the number of germ cells reaching a maximum of about 7 million at mid gestation. Although mitosis can continue until birth, by the third trimester cell loss exceeds the rate of mitosis, and the number of germ cells falls. This loss occurs by apoptosis, or programmed cell death, and is mainly seen within the sex cords in oogonia and oocytes that have not formed follicles by association with somatic cells.

Proliferation of the coelomic epithelium forms protrusions into the mesenchyme, which gives rise to the primary sex cords, surrounding groups of primordial germ cells/oogonia. Shortly after, outgrowths of cells from the mesonephros form primordial sex cords containing the germ cells. At mid gestation, the sex cords intermingle and account for approximately 60% of the ovarian volume, falling to 14% by birth while the medulla of the ovary relatively enlarges. Meanwhile, oogonia begin to cease mitosis and enter meiosis at 10 to 12 weeks after conception, some weeks after sex-specific gonadal differentiation. This division is arrested 1 to 2 weeks later, at diplotene of the first meiotic division, resulting in the formation of oocytes. The oocyte remains arrested in the first meiotic division unless and until the follicle reaches the mature antral stage and is subjected to the gonadotropin surge preceding ovulation, which may be many decades later.

Newly formed oocytes become enclosed in a single flattened layer of somatic pregranulosa cells resting on a basement membrane, to form the primordial follicle. Follicle formation begins close to the corticomedullary boundary, and primordial follicles appear to separate from the sex cords. The origin of granulosa cells is still not completely certain, and may vary from one species to another, but it is likely that they are derived from the ovarian surface epithelium.

Folliculogenesis in the normal ovary

Primordial follicles provide the stock of oocytes, which must last for up to 50 years. Initiation of follicle growth, i.e. progression of the follicle from the primordial to the early growth phase, must be tightly regulated to ensure a steady supply of oocytes for ovulation during a normal reproductive lifetime. However, the factors responsible for controlling the initiation of growth remain to be elucidated. The first indication of growth of the follicle is a change in shape of the granulosa cells, which become more cuboidal in appearance. Follicles then pass through a transitional or intermediary stage, in which a proportion of the granulosa cells are cuboidal and the rest remain flattened. This is followed by the primary stage, in which the oocyte is enclosed in a single layer of completely cuboidal cells. By this stage the oocyte has increased significantly in volume. Follicle development progresses by formation of a second layer of granulosa cells, and at this stage the first theca cells, derived from surrounding stroma, begin to organize around the granulosa layer. This is followed by the formation of further layers of granulosa and theca cells (with enlargement of the oocyte) to form a multilayered preantral follicle. The outer layers of the theca comprise cells that are similar to those in surrounding stroma, and constitute the theca externa. The cells of the inner layers become polyhedral and form the theca interna, the site of androgen production in large preantral and antral follicles. Eventually, the theca interna of each follicle receives its own blood supply. Development of the follicle to the multilayered preantral stage can progress without the need for gonadotropins. It is unclear how long it takes for a follicle to progress from the primordial stage to a large preantral follicle, but estimates suggest that it may be at least 6 months.

Granulosa cells continue to proliferate, and a fluid-filled space (the antrum) eventually forms between them, and continues to enlarge. The follicle is now an antral follicle, the stage at which the endocrine system comes into play and gondaotropins take over control of folliculogenesis. It is from this stage that the biggest expansion of the follicle occurs, in terms of granulosa and theca cell numbers, antrum size, oocyte growth, and overall follicle diameter. Follicles that reach this stage are considered to be part of a selectable pool of follicles from which the dominant follicle will arise (i.e. the one most likely to complete maturation and ovulate). This pool may number approximately 15 to 20 follicles between the 2 ovaries in young women, and declines with age, averaging 10 at 30 years and 5 at 40 years. It will be evident that only a small fraction of the total pool of follicles is destined to ovulate. The rest will undergo atresia (death by apoptosis). Although it is likely that follicle loss by atresia occurs at all stages of folliculogenesis, the highest proportion of atretic follicles is seen during the gonadotopin-dependent antral stages. As described below, selection of a single follicle for ovulation in the human menstrual cycle inevitably involves regression and demise of subsidiary follicles within the same cohort.

The hypothalamic-pituitary-ovarian axis

Like the testis, the ovary has two major functions: (1), the production of gametes and (2), the secretion of hormones (particularly sex steroids) that affect development and function of the reproductive tract, as well as having important peripheral affects on muscle, bone, and skin. Like all classic endocrine organs, the function of the ovary is dependent upon regulation by pituitary hormones, which in turn are regulated by hypothalamic signals.

Gonadotropin-releasing hormone (GnRH) is a decapeptide secreted by the hypothalamus in a pulsatile manner, the frequency of pulses (between 60 and 180 min, according to the stage of the menstrual cycle) having a profound influence on the response of the pituitary gonadotropins to GnRH. The episodic secretion of GnRH is reflected in the pattern of circulating gonadotropins, the pulses of luteinizing hormone (LH) being more discrete than those of follicle-stimulating hormone (FSH) because of the shorter half-life of LH in the circulation. LH and FSH, which are glycoproteins, act in concert on the maturing large ovarian follicles: LH stimulates the thecal layer of the follicle to produce androgens (androstenedione and testosterone), whereas FSH acts specifically on the inner, granulosa cell layer of the mature follicle, which lacks the capacity to synthesize androgens, to convert androgens to oestrogens—the so-called ‘two cell, two gonadotropin’ hypothesis. Following ovulation, oestradiol continues to be produced by the corpus luteum, but the principal circulating steroid at this stage of the cycle is progesterone. Oestradiol, during the mid follicular phase of the cycle (see ‘The menstrual cycle’, below), and progesterone (luteal phase) exert negative feedback on both the pituitary and the hypothalamus to inhibit the secretion of gonadotropins. The ovary also produces two closely related nonsteroidal glycopeptide hormones that selectively inhibit FSH and contribute to the negative feedback: inhibin B (produced by developing follicles in the follicular phase) and inhibin A (produced mainly by the corpus luteum). The extraordinary feature of the hypothalamic–pituitary–ovarian axis is the phenomenon of positive-feedback stimulation of gonadotropins by oestradiol in mid cycle, which results in the LH surge and ovulation, as described below.

The menstrual cycle

At the beginning of each normal menstrual cycle (conventionally taken as the first day of menses), there is a cohort of follicles, ranging between 2 and 5 mm in diameter, that are dependent on and responsive to FSH. Between the late luteal phase of the previous cycle and the early follicular phase, the negative-feedback signal primarily provided by progesterone is removed, and the concentration of FSH rises. This intercycle increase in FSH exceeds a notional threshold level that encourages follicle maturation. Of the cohort of follicles that arrive at this FSH threshold, only one (or occasionally two) is destined to complete the journey to ovulation. This is the follicle that is most responsive to FSH. It is often the largest of the cohort, but not necessarily so. As the follicles grow in response to FSH, oestradiol (and inhibin B) levels rise in the circulation and exert a negative-feedback effect on FSH. As a result of the fall in FSH in the mid follicular phase, most of the follicles in the cohort will regress and die by atresia, leaving only the most FSH-sensitive, dominant follicle to continue to grow and to secrete oestradiol. By this time the granulosa cells of the dominant follicle have acquired LH receptors (the only time in the life of the follicle when this occurs) and are thus now responsive to LH. The dual effect of FSH and LH enhances granulosa cell differentiation and steroidogenesis, so that in the preovulatory phase of the cycle serum oestradiol levels in the circulation have increased more than 10-fold compared with the early follicular phase, 95% of circulating oestradiol being attributable to that single preovulatory follicle. The steeply rising levels of oestradiol (probably assisted by a small increase in circulating progesterone, signalling granulosa cell and perhaps oocyte maturation) then trigger the LH surge; the only example of a positive-feedback effect of a target hormone on the hypothalamic–pituitary unit. The LH surge has three main functions: (1), it signals resumption of meiosis in the oocyte ready for fertilization, (2), it leads to follicle rupture and ovulation, and (3), it stimulates formation of the corpus luteum, converting the follicle from a mainly oestrogen-producing unit to a highly vascularized progesterone factory. Progesterone suppresses gonadotropins during the luteal phase and, if conception does not occur, luteolysis ensues after an apparently preprogrammed interval of 12 to 14 days, triggering the onset of a new cycle.

Disorders of ovulation

Clinical presentation and causes of anovulation

Disorders of ovulation usually result in perturbation of normal cyclical menses. It is uncommon to have regular but anovulatory cycles, the exception being during adolescence, when cyclical ovarian activity without ovulation (a feature of immaturity of the hypothalamic–pituitary–ovarian axis) is characteristic of the early months after menarche. Thus, anovulation is generally characterized by amenorrhoea, oligomenorrhoea (>6 weeks between periods), or very irregular menses. Amenorrhoea may be primary (i.e. no previous periods) or secondary (at least one previous spontaneous period). Primary amenorrhoea is less common, and although its causes overlap with those of secondary amenorrhoea, it is not surprising that disorders of development of the ovaries and/or reproductive tract are overrepresented in this category. In some cases, primary amenorrhoea (defined as menarche delayed beyond 16 years of age) is accompanied by delayed pubertal development. Menstrual disturbance may be accompanied by symptoms of oestrogen deficiency, including vaginal dryness and hot flushes. Interestingly, vasomotor symptoms are common in women with primary ovarian failure, but not in those with oestrogen deficiency resulting from hypothalamic–pituitary dysfunction. Patients with hyperprolactinaemia may report inappropriate lactation (galactorrhoea), but it is important to recognize that this affects only 30 to 50% of women with hypersecretion of prolactin. Menstrual abnormalities accompanied by symptoms of androgen excess (hirsutism, acne, or alopecia) are typical of polycystic ovary syndrome (PCOS). Examination should include routine measurement of height and weight, and calculation of body mass index (BMI).

Bullet list 1 Causes of secondary amenorrhoea

The causes of secondary amenorrhoea are summarized in Bullet list 1 above.

The most prevalent cause of secondary amenorrhoea is hypothalamic and/or pituitary dysfunction. Hyperprolactinaemia, weight loss-related, and idiopathic amenorrhoea are all associated with a functional rather than structural hypothalamic disorder of gonadotropin regulation (see below). PCOS accounts for a further 32% of cases, and primary ovarian failure for 11%. Among women presenting with oligomenorrhoea , the underlying cause is PCOS in the great majority of cases, making this the most common overall cause of anovulation, as discussed in more detail below.

Differential diagnosis of amenorrhoea and oligomenorrhoea

With the aid of a small number of endocrine investigations it is possible clearly to differentiate between the various causes of ovarian disorders, and to guide management (Table 1). Measurement of serum FSH will distinguish primary ovarian failure (wherein FSH is elevated) from other causes of amenorrhoea and oligomenorrhoea (in which FSH is normal or low). Assessment of oestrogen status is an important step in the investigation of women with amenorrhoea. This can be achieved by direct measurement of serum oestradiol, by ultrasonographic measurement of endometrial thickness, or by a progestagen challenge test. Serum oestradiol measurements are valuable if results are unequivocally low (i.e. lower than in the early follicular phase) or normal (equivalent to mid follicular phase levels), but concentrations in the early follicular phase range may not exclude chronic oestrogen deficiency. The advantage of ultrasonography or the progestagen challenge test is that these provide what amounts to an in vivo bioassay of endogenous oestrogen action (on the endometrium).

A combination of low oestrogen and low (or normal) FSH is indicative of hypothalamic/pituitary dysfunction. In such cases serum prolactin should be measured and, if it is elevated, pituitary imaging performed (see below).

Bullet list 2 Causes of oligomenorrhoea

Overview of management of disorders of ovulation

A simple schema for the differential diagnosis of ovarian disorders provides a basis for the selection of appropriate treatment, as outlined in Table 1. Details of management of the individual disorders are given in the appropriate sections below. The first principle must always be to treat any underlying cause, if possible, e.g. helping women with weight loss-related amenorrhoea to gain weight. For primary ovarian failure (high FSH, low oestrogen), gamete donation is the only option for fertility treatment, but oestrogen/progestagen is required for the treatment of symptoms of oestrogen deficiency (see below). In women with a hypothalamic or pituitary cause of anovulation (low or normal FSH with low oestrogen), ovulation can be induced by gonadotropins or GnRH (or, in the case of hyperprolactinaemia, by dopamine agonists), but patients not requiring fertility treatment will, like those with primary ovarian failure, need sex-hormone replacement. Ovulation can be induced by antioestrogens in most women with PCOS (normal FSH, normal oestrogen), but some may require gonadotropin therapy. In those not wishing to conceive, management of erratic periods and treatment of attendant symptoms of androgen excess are important considerations.

Premature (primary) ovarian failure

Ovarian failure usually occurs after the end of the fifth decade and manifests as the menopause, defined as the last ever menstrual period (average age 51 years). It is described as being premature if it occurs under the age of 40 years, and this affects approximately 1% of women. Unfortunately, regardless of the cause, premature ovarian failure is irreversible, although it is not unusual to have episodes of spontaneous ovarian function after the menopause. In a very small proportion (1–2%) of women this may even result in pregnancy. The characteristic endocrine features are oestrogen deficiency associated with elevated serum concentrations of FSH (and LH), sometimes referred to as hypergonadotrophic amenorrhoea.

Types

Idiopathic

Most cases of primary ovarian failure are idiopathic, and it is not clear if the underlying cause is related to an initially reduced population of primordial follicles within the fetal ovary, an increased rate of atresia throughout reproductive life, or a combination of both. It may be associated with autoimmune conditions and may be accompanied by ovarian, thyroid, adrenal, and antiendomysial autoantibodies. There is also an association with type 1 diabetes. The significance of ovarian autoantibodies in the aetiology of ovarian failure remains uncertain, however. ‘Resistant ovary syndrome’ refers to a state in which oestrogen-deficient amenorrhoea or oligomenorrhoea is associated with high serum FSH, but with the persistence of follicles in the ovary. This is now generally recognized as a stage of ovarian failure, inevitably culminating in further reduction and, finally, exhaustion of the follicle pool.

Chromosomal and genetic

It is important to investigate potential causes of premature ovarian failure, as these may be significant in counselling family members or any children of affected women. The commonest forms of gonadal dysgenesis are Turner’s syndrome (45,X) and its mosaic forms. These women have a structurally normal vulva, vagina, uterus, fallopian tubes, and ovaries, although the latter may be small and will undergo premature failure. An oocyte requires an X chromosome to protect it from early atresia, and women with Turner’s syndrome or mosaic Turner’s syndrome are highly likely to experience premature ovarian failure. Indeed, many of these women will have primary amenorrhoea.

Other causes of gonadal dysgenesis include Swyer’s syndrome (46, XY gonadal dysgenesis) which invariably results in delayed puberty and primary amenorrhoea. Approximately 30% of cases are caused by a deletion on the Y chromosome that results in nonfunctioning streak gonads, which are unable to secrete antimullerian hormone (normally a product of Sertoli cells in the fetal testis) and therefore the mullerian structures do not regress. The result is an anatomically normal vulva, vagina, uterus, and fallopian tubes. Mutations on autosomal chromosomes have also been described and although cases may be familial, the majority are caused by new mutations. The gonads are thought to be at increased risk of malignant change, and surgical removal is usually recommended. The gonads often lie high on the pelvic side-wall, or may even be found above the pelvis, and can be removed laparoscopically.

Autosomal dominant, autosomal recessive, and X-linked patterns of inheritance have also been described in primary ovarian failure. It is now recognized that the mutation responsible for fragile X syndrome, a cause of learning difficulties, can be associated with premature ovarian failure in carriers.The syndrome is caused by a full mutation in the FMR1 gene and any of its unstable premutations can be associated with ovarian failure. These premutations are probably the commonest known genetic cause of ovarian failure.

Iatrogenic

Iatrogenic premature ovarian failure is increasingly seen in young adult survivors of childhood malignancies, especially haematological ones. The cause is loss of germ cells and follicles as a consequence of chemotherapy and/or irradiation. A second group seen with increasing frequency is young women who have had adjuvant chemotherapy for breast cancer. Not all chemotherapy regimens are equally toxic to the ovary and the effects are variable. Alkylating agents, especially cyclophosphamide and procarbazine, are particularly associated with reduced ovarian function and premature ovarian failure, although platinum-based agents are less so.

Treatment

Treatment of primary ovarian failure falls into four categories: (1), induction of puberty (using low-dose oestrogen) in girls with delayed puberty, (2), control of symptoms of oestrogen deficiency, (3), preservation (or improvement) of bone mineral density, and (4), improvement of fertility. Hypo-oestrogenic symptoms include hot flushes and night sweats, dyspareunia, urinary frequency, and loss of libido. There is also often a loss of a sense of general well-being, although that may be due in part to a reaction to the diagnosis. Bone mineral density is likely to be low in untreated ovarian failure, but can be preserved and increased with exogenous oestradiol. Oestrogen replacement is usually in the form of a sequential or continuous combined regimen, progesterone being required to protect the endometrium from the effects of unopposed oestrogen.

Some young women find the combined oral contraceptive pill a more acceptable form of oestrogen replacement. Compliance can be an issue, especially in those under 25 years, and the pill can be a useful alternative to conventional hormone replacement. Another advantage of relying on the pill for oestrogen replacement is that it is contraceptive, whereas conventional hormone replacement therapy is not. Although pregnancy after premature ovarian failure is uncommon, it can and does occur. Unplanned pregnancy for any woman can be traumatic, but when it occurs in one affected by premature ovarian failure the results can be particularly upsetting. An important part of the management is therefore contraceptive advice, if pregnancy is undesirable. By contrast, a woman who would welcome a pregnancy can be reassured that hormone replacement treatment will not decrease her chances of a pregnancy occurring.

Fertility

Once premature ovarian failure has occurred no fertility treatment will assist conception, and oocyte donation is the only option. It is an extremely successful treatment for many couples, but the main difficulty is finding an appropriate donor. Pelvic irradiation is associated with a poor outcome from oocyte donation because of attendant uterine abnormalities. The endometrium may be unable to support implantation, and even if pregnancy occurs the obstetric risks are increased, including miscarriage, premature delivery, and intrauterine growth restriction. The risks of pregnancy must also be seriously considered for women with Turner’s syndrome, as maternal deaths from aortic arch dissection have occurred, and there appears to be an increased incidence of placental abruption.

Hypothalamic/pituitary dysfunction

Weight-loss related amenorrhoea

The majority of cases of amenorrhoea resulting from hypothalamic/pituitary dysfunction are of hypothalamic rather than pituitary origin, and most are a function of an underlying disorder. Weight loss-related amenorrhoea is very common. Nutritional status is an important determinant of reproductive function, and being underweight (BMI <19 kg/m²) is very likely to result in abnormalities in the pulsatile secretion of GnRH. This in turn leads to reduced frequency and amplitude of LH and FSH pulses, and oestrogen-deficient amenorrhoea. Cyclical ovarian function can be restored by weight gain, but this is not usually easy to effect. Most women with amenorrhoea related to weight loss have an underlying eating disorder and it is often necessary to seek the help of specialist psychological or psychiatric services. Fertility treatment is unwise in underweight women until they have reached a normal BMI. Although it is possible to induce ovulation with GnRH or gonadotropins, underweight women are at considerably increased risk of having a small baby. However, correction of oestrogen deficiency is appropriate while treatment to aid weight gain is under way.

Hyperprolactinaemia

Hyperprolactinaemia is another common cause of oestrogen-deficient amenorrhoea. It is discussed in greater detail in Chapter 13.2. Amenorrhoea is the typical presenting symptom of hyperprolactinaemia. It is important to exclude primary hypothyroidism or concurrent medication as possible causes before embarking upon pituitary radiology. In particular, dopamine antagonists (e.g. phenothiazines and metoclopramide) are well-recognized causes of elevated serum prolactin. MRI is the preferred method of detecting pituitary abnormalities. A microadenoma of the pituitary may be found in up to 50% of women with hyperprolactinaemic amenorrhoea. Larger tumours (≥10 mm) are much less common. Management of hyperprolactinaemic amenorrhoea, even in women with an obvious prolactinoma, is primarily by the use of long-acting dopamine agonists such as bromocriptine or cabergoline. These drugs lower prolactin, restore ovulatory function, and typically reduce the size of prolactin-secreting tumours. Pituitary surgery is rarely needed, even in women with large prolactinomas.

Idiopathic hypothalamic amenorrhoea

In about 10% of cases the underlying cause of hypothalamic amenorrhoea is uncertain. Recent studies of women with idiopathic (functional) hypothalamic amenorrhoea have suggested that this category of patients represent what is essentially a stress-related hypothalamic disorder. Such patients respond very well to cognitive behavioural therapy, which results in resumption of ovulatory cycles without the need for endocrine treatment. If cognitive behavioural therapy is unsuccessful, ovulation can be induced by GnRH or gonadotropins in women seeking fertility treatment. Otherwise, oestrogen/progestagen replacement is desirable to treat symptoms of oestrogen deficiency and/or to maintain bone density. Finally, it is important to recognize that other hypothalamic–pituitary disorders, although themselves being rare causes of amenorrhoea, may first present with menstrual dysfunction. Congenital deficiency of GnRH, best illustrated by Kallmann’s syndrome (in which gonadotropin deficiency is associated with anosmia), often presents as delayed puberty, but may manifest as primary amenorrhoea in girls who have completed pubertal development. Amenorrhoea is a common presenting symptom in women with acromegaly or Cushing’s syndrome. Hypothalamic tumours or granulomas may cause deficiency of not only GnRH but also other hypothalamic hormones. It is not necessary routinely to screen for these rarer causes of amenorrhoea, but it is important to be alert to features in the history and examination that may suggest a more unusual diagnosis.

Polycystic ovarian syndrome (PCOS)

PCOS is the commonest of all the ovarian disorders and the commonest endocrine disorder in women of reproductive age, with a prevalence greater than 5% of the general population. The typical clinical presentation is the association of features of anovulation or oligo-ovulation (amenorrhoea or menstrual irregularity) with clinical and/or biochemical evidence of androgen excess (hirsutism, persistent acne, or male-pattern alopecia) in women with polycystic ovaries. However, the recognition that there may be a broader spectrum of clinical and biochemical presentation has led to a recent revision of the diagnostic criteria for PCOS (see below). The aetiology of PCOS remains uncertain. There is strong evidence for an ovarian origin of androgen excess, although the hypersecretion of adrenal androgens can also be found, albeit in a minority of patients with PCOS. Genetic factors clearly play a part in the aetiology; there is clustering of cases of PCOS within families, and a recent twin study showed that the concordance of features of PCOS is significantly greater in identical than in nonidentical twins. The mode of inheritance is unclear, but it is unlikely to be a simple mendelian trait. Rather, like type 2 diabetes, it is a complex endocrine disorder in which several genes may play a part. In addition, as in type 2 diabetes, the phenotype is modified by environmental factors, and obesity clearly exacerbates endocrine and metabolic dysfunction, and is associated with more severe symptoms.

Definition and diagnostic criteria

The classic definition of PCOS, i.e. hyperandrogenism associated with chronic anovulation (in the absence of any confounding pituitary or adrenal disorders), is notable for its lack of reference to ovarian morphology, and yet almost all women who meet these criteria will have polycystic ovaries. In addition, polycystic ovaries can be found in women with symptoms of hyperandrogenism, but who have regular menstrual cycles, as well as in those with anovulation but no evidence of androgen excess. A consensus meeting held in 2003 revised the diagnostic criteria, allowing a more inclusive definition (Table 2). This revision has inevitably led to some controversy about the definition of PCOS, but there is ample evidence that women with polycystic ovaries who present with hyperandrogenism, but have regular cycles, and those with oestrogen-replete amenorrhoea or oligomenorrhoea, but who have no features of androgen excess, simply have varying forms of the same underlying condition.

Endocrine features

The heterogeneity of the clinical features of PCOS extends to the endocrine abnormalities associated with it As a result, specific endocrine parameters are not a requirement for diagnosis, although measurement can be helpful to support it and, importantly, to exclude other conditions. Until the advent of widely available high-resolution ultrasonography, the diagnosis of PCOS was usually based on a combination of biochemical and clinical features.

Raised serum testosterone concentration is the most common biochemical abnormality in PCOS, occurring in about 70% of cases. The free androgen index, calculated from total testosterone and sex-hormone binding globulin (SHBG), has been found to be a useful marker by some clinicians. However, since SHBG is closely associated with BMI and, more particularly, abdominal circumference, the increased free androgen index found in PCOS is at least in part a reflection of increased abdominal adiposity. Serum concentrations of the weak androgen androstenedione are also elevated in PCOS. In practical terms, measuring serum testosterone is usually preferable to measuring androstenedione, as the process is automated in most clinical laboratories and therefore more cost-efficient. In 10 to 20% of patients with PCOS, serum levels of the weak adrenal androgen dehydroepiandrosterone sulphate (DHEAS) are also modestly elevated, suggesting that, at least in some patients with PCOS, there may be an adrenal contribution to increased circulating androgens.

Among women with PCOS, clinical signs of hyperandrogenaemia (e.g. hirsutism) are associated with higher testosterone levels than in those without. The presence or absence of features of hyperandrogenism, however, does not accurately predict serum androgen levels, as clinical expression depends on the peripheral conversion of testosterone to its active metabolite 5α-dihydrotestosterone by 5α-reductase, as well as on end-organ sensitivity (androgen receptor activity) (see ‘Other causes of hyperandrogenism in women’, below). Obesity in PCOS is associated with higher free testosterone levels than in lean counterparts, and in part reflects the lower SHBG levels found in the former group. In addition, obesity may have an independent effect on peripheral androgen metabolism since androsterone glucuronide levels, a marker for peripheral 5α-reductase activity, are raised in this group. Genetic factors may affect end-organ sensitivity, e.g. PCOS occurs in Chinese and Japanese women, but hirsutism is relatively uncommon in these populations. By contrast, hirsutism features commonly in women with PCOS from the Indian subcontinent.

Women with PCOS tend to have higher LH levels than those with normal ovaries. The highest prevalence of elevated LH levels is in those with anovulatory menses or amenorrhoea, but even in this group more than 40% will have normal LH. By contrast, FSH levels are normal but tend to be lower than in the normal early follicular phase. Many have cited a raised LH:FSH ratio (either 2.5:1 or 3:1) as a diagnostic feature of PCOS, but it is neither sensitive nor specific enough to be used as a reliable diagnostic criterion.

Oestrogen levels in women with all variants of PCOS are normal. As discussed previously, this can be used to distinguish between oligo/amenorrhoeic women with PCOS and those with other causes of anovulation such as hypothalamic or pituitary disorders, or ovarian failure. Plasma oestradiol levels in PCOS lie within the range normally seen in the early to mid follicular phase of the menstrual cycle, but oestrone levels are significantly higher. This is probably because of the increased peripheral conversion of high levels of circulating androstenedione to oestrone in adipose tissue.

Hyperprolactinaemia has been described in association with PCOS, but this usually reflects spurious fluctuations in serum prolactin; it probably occurs no more commonly than in the normal population and is rarely a persistent problem.

Metabolic abnormalities

PCOS is not just a reproductive disorder, it is also associated with a characteristic metabolic abnormality, central to which are peripheral insulin resistance and compensatory hyperinsulinaemia. Insulin resistance is independent of body weight, but the difference between women with PCOS and controls is amplified with increasing body weight. Reduced insulin sensitivity is related to an abnormality in energy balance, specifically, reduced postprandial thermogenesis, which may contribute to the development of obesity. Interestingly, insulin resistance in PCOS appears to be confined to (or at least is most apparent in) the major subgroup of women who have both anovulation and hyperandrogenism (Bullet list 3).

Typically, women with PCOS have increased abdominal adiposity and visceral fat accumulation, and this is correlated with insulin resistance. There is also an associated dyslipidaemia, characterized by lower than normal serum concentrations of high-density lipoprotein cholesterol and elevated levels of low-density lipoprotein cholesterol. Although glucose tolerance is often normal in these women, impaired glucose tolerance has been noted in 10 to 40% of young obese women with PCOS. The defect in insulin action associated with PCOS appears to be secondary to a defect in postreceptor signal transduction, and shows subtle differences from that found in other insulin-resistant states. The major defect associated with PCOS, independent of obesity, is in insulin signalling in classic insulin target tissues such as muscle. Suppression of hepatic gluconeogenesis is reduced, but only in obese PCOS women. By contrast, obesity alone has a smaller effect on the sensitivity of insulin-mediated glucose utilization, but a greater effect on the rate of glucose utilization.

There is some debate as to whether the insulin resistance in PCOS represents a primary defect in insulin action, or whether it is secondary to hyperandrogenism and/or the result of increased truncal–abdominal fat. The interaction of insulin and androgens is complex. Experimental data suggest that androgens affect the flux of free fatty acids from visceral fat deposits, which may in turn affect insulin sensitivity. However, therapeutic reduction of serum androgen levels does not appear to improve insulin sensitivity. On the other hand, hyperinsulinaemia clearly affects androgen production. Insulin has gonadotropic activity and can influence ovarian steroidogenesis by both theca and granulosa cells via an interaction with LH. In addition, hyperinsulinaemia reduces the hepatic production of SHBG and thereby raises levels of non-protein-bound (i.e. biologically available) testosterone.

Bullet list 3 Typical metabolic features of PCOS

Reproductive consequences

PCOS is by far the commonest cause of anovulatory infertility, accounting for more than 75% of cases. Anovulation is undoubtedly the principal reason for subfertility in women with polycystic ovaries, but there has been some speculation that polycystic ovaries, in the absence of the syndrome, may contribute to problems with fertility. Polycystic ovaries are found more commonly than in the general population in infertile ovulatory women with tubal disease (50%), in women whose partners have sperm dysfunction (53%), and in couples with unexplained infertility (44%). Women with polycystic ovaries are also over-represented among women with a history of recurrent miscarriage (three or more consecutive miscarriages). However, the live birth rate of ovulatory women with polycystic ovaries, after spontaneous conception, is the same as that in a well-matched population of women with normal ovaries.

Long-term consequences

The significance of PCOS for women’s health at a population level is increasingly being recognized. Although management of symptoms such as infertility and hirsutism is important, consideration must also be given to management and, if possible, prevention of the long-term effects of the disorder. These include an increased risk of developing endometrial cancer and the consequences of metabolic abnormalities, namely diabetes and cardiovascular disease.

Endometrial carcinoma

PCOS has been recognized as a risk factor for endometrial carcinoma since the 1950s, and there are reports of the disease occurring in young (premenopausal) women with PCOS. In women with PCOS who have amenorrhoea or infrequent menses, the endometrium is exposed to prolonged stimulation with oestrogen in the absence of cyclical progesterone (unopposed oestrogen). This may lead to endometrial hyperplasia and, if untreated, to endometrial carcinoma. Obesity adds to the risk of developing endometrial cancer by a number of interrelated intermediary factors, including increased oestrogen production, hyperinsulinaemia, and reduced serum SHBG.

Gestational diabetes

The link between PCOS, insulin resistance, and impaired glucose tolerance suggests that women with PCOS are at increased risk of developing both type 2 diabetes and gestational diabetes. The physiological insulin resistance of pregnancy is added to that intrinsic to PCOS, and may unmask impaired pancreatic β-cell function. The evidence for an increased risk of gestational diabetes among women with PCOS is suggestive, but not yet compelling. Most of the studies to date have been small and retrospective, and involve ethnically mixed populations. However, a recent meta-analysis of the available data suggests a threefold increase in the risk of gestational diabetes in women with PCOS.

Type 2 diabetes

As indicated above, impaired glucose tolerance and even frank diabetes are common in obese young women with PCOS. Longitudinal studies have been limited, both in number and in duration of follow-up, but those that are available indicate that the prevalence of both impaired glucose tolerance and diabetes increase, as might be predicted, with age and, inevitably, BMI. Likewise, population studies have been few, but the results support the view that PCOS is a significant risk factor for the development of type 2 diabetes. The relative risk is around twofold after adjustment for obesity, but rises to three- to sevenfold in obese women with PCOS.

Cardiovascular risk

PCOS is associated with well-recognized risk factors for cardiovascular disease, namely obesity, insulin resistance, dyslipidaemia, diabetes, and (in some but not all studies) hypertension. In addition, surrogate markers of cardiovascular disease have also been found to be abnormal. Endothelial function is impaired in young women with PCOS. Carotid artery intima–media wall thickness (associated with an adverse cardiovascular risk profile in middle-aged and older general populations) is increased in women with PCOS over the age of 45 years, and carotid plaques are more common. Coronary artery calcification is a marker for coronary atherosclerosis, and is also more common in women with PCOS than BMI-matched controls. Left ventricular mass index was found to be increased, and diastolic dysfunction present in obese and nonobese young women with PCOS, suggesting a detrimental effect on the cardiovascular system, although this is yet to be confirmed.

It might be expected from the presence of multiple risk factors for cardiovascular disease that women with PCOS, especially if obese, would have an increased morbidity and mortality from the condition. There are few epidemiological studies and no substantial longitudinal studies, but the data so far suggest that there are fewer cardiovascular events than would be predicted from the cluster of risk factors. The two largest studies give a similar odds ratio (1.5) for the risk of cardiovascular events. In both studies the populations were under 60 years of age, so it remains possible that the relative risk of heart attack (and stroke) will increase with age. An alternative explanation is that there are factors in women with PCOS that are protective against cardiovascular disease, e.g. as a result of unopposed oestrogen or even raised androgen levels. Perhaps most importantly of all, it must be appreciated that the presence of risk factors does not prove the presence of the disease, and that surrogate markers are not necessarily reliable predictors of outcome.

Management of PCOS

The management of PCOS is mainly targeted at the relief of symptoms. Symptoms of androgen excess, including hirsutism, acne, or alopecia, can be attenuated by the use of antiandrogens such as cyproterone acetate and spironolactone (which, in the absence of cyproterone acetate, is widely used in the United States of America). Flutamide is a pure antiandrogen (unlike cyproterone acetate it has no progestagenic activity), but its place is less secure in the management of symptoms of androgen excess because there are fewer studies to support its routine use and there have been reports of hepatic toxicity. Low-dose cyproterone acetate may be conveniently combined with ethinylestradiol (as co-cyprindiol) and this preparation is particularly useful in women with accompanying menstrual disturbance. It is also an effective contraceptive. For those in whom oestrogen is contraindicated cyproterone acetate can be given alone, but nonhormonal contraception should be advised in those at risk of pregnancy because of the theoretical risk of feminization of a male fetus. Women often seek medical help with hirsutism when beauty treatments such as waxing, plucking, electrolysis, and laser hair removal become inconvenient or too expensive. It is important to ensure a realistic expectation of treatment, which is that antiandrogens should be used as adjuncts, not replacements, to beauty treatments. In addition, hormone treatment may take 9 to 12 months to become fully effective; an improvement in hirsutism is often best judged by a reduced frequency of hair-removal treatment. Antiandrogen therapy is also effective for acne but, unfortunately, alopecia rarely improves with antiandrogen treatment, and the objective here is to limit further hair loss. It is therefore important to treat early signs of androgen-dependent hair loss.

Anovulatory women with PCOS who wish to conceive usually respond to ovulation induction therapy. The principle is to raise serum FSH levels to encourage development of a single, healthy, dominant follicle (and therefore limit the risk of multiple pregnancy). The first-line treatment is the antioestrogen clomifene, to which 75 to 80% of women will ovulate in response. In those who do not respond or conceive after six or more ovulatory cycles, treatment with exogenous gonadotropin is appropriate. The modern approach is to start with a low dose of FSH and if necessary make small increments in dose to find the threshold for development of a single dominant follicle. Even low-dose FSH treatment requires close monitoring to reduce the risks of multiple pregnancy. An alternative to gonadotropin treatment is laparoscopic ovarian diathermy, a single, if invasive, procedure. However, surgery alone will result in ovulatory cycles in less than 50% of subjects, and adjuvant clomifene or FSH treatment is often required.

In anovulatory women with PCOS who do not wish to conceive, regulation of menses can be ensured by treatment with a combined oral contraceptive or cyclical progestagen treatment. Because of the risk of endometrial hyperplasia or cancer it is important to offer such treatment, even in women with amenorrhoea or oligomenorrhoea who are not concerned about lack of periods.

In obese women with PCOS calorie restriction is not only desirable, but also surprisingly effective in improving the symptoms of PCOS, particularly menstrual pattern and fertility. Dietary restriction leading to merely a 5 to 10% reduction in weight is associated with much improved ovarian function. Overweight and obese women respond poorly to induction of ovulation, and from an obstetric viewpoint the risks of gestational diabetes and pregnancy-related hypertension are increased. Weight reduction before fertility treatment, though never easy to achieve, is therefore an important aspect of management. Evidence from the Diabetes Prevention Program, a prospective study of men and women with impaired glucose tolerance, suggests that calorie restriction coupled with lifestyle changes (including increased exercise) will reduce the risk of conversion to diabetes. Although there are, as yet, no such studies in women with PCOS, it is logical that such an approach will also reduce the chance of developing diabetes in this at risk group.

The Diabetes Prevention Program also showed that the biguanide metformin, which has long been used for the treatment of type 2 diabetes, was effective in reducing conversion from impaired glucose tolerance to diabetes (although significantly less so than diet and lifestyle changes). In recent years metformin has been enthusiastically advocated for the management of PCOS, even in the absence of impaired glucose tolerance. A large number of publications have supported its use in fertility treatment (particularly in combination with clomifene), menstrual regulation, and management of hirsutism. However, there have been few large randomized controlled trials of metformin in the management of PCOS, and those few adequately powered studies that have been performed to date have failed to support those claims. It remains to be seen whether metformin has a role in diabetes prevention in women with PCOS.

Other causes of hyperandrogenism in women

Hyperandrogenism, in this context defined as clinical evidence of androgen excess in women, is a common and distressing problem. Hyperandrogenism manifests itself as hirsutism, persistent acne, or androgenic alopecia. Although PCOS is the commonest cause of androgen excess, it is important to consider other possible diagnoses.

During puberty circulating androgen concentrations rise, and the familiar pattern of androgen-dependent body (terminal) hair growth is seen. In normal premenopausal women the adrenal is the predominant source of androgens. Testosterone is the most important circulating androgen, and is secreted by both ovaries and adrenals. But about 50% of circulating testosterone is derived by conversion from androstenedione (a weak androgen) in peripheral tissues such as skin and adipose. More than 90% of circulating testosterone is bound either to SHBG or albumin. Only the unbound (and possibly albumin-bound) testosterone is available to target tissues. Testosterone is further metabolized within the hair follicle to the more potent androgen dihydrotestosterone by the enzyme 5α-reductase. Both testosterone and (with a higher affinity) dihydrotestosterone bind to specific androgen receptors within the hair follicle to affect the growth of terminal hair. The biological effect of androgens may also be regulated at the level of the androgen receptor itself. Recent evidence suggests that heterogeneity of the androgen receptor is conferred by epigenetic modification of the androgen-receptor gene, and that these modifications are related to clinical indices of androgenicity.

Causes of hirsutism

The causes of hirsutism are summarized in Bullet list 4.

Hirsutism is most commonly caused by PCOS, which accounts for about 90% of cases, including those who might previously have been labelled as having idiopathic hirsutism. However, hirsutism may be a manifestation of other, much rarer but more serious endocrine disorders, such as Cushing’s syndrome and adrenal or ovarian tumours. Careful clinical evaluation is the key to differential diagnosis. Long-standing mild to moderate hirsutism, with or without menstrual disturbance, is suggestive of PCOS or idiopathic hirsutism, whereas a short history of increasing hirsutism in a previously nonhirsute subject should alert the physician to the possibility of an alternative diagnosis. Hirsutism and menstrual disturbances are common presenting features in women with Cushing’s syndrome or androgen-secreting tumours. In the case of Cushing’s syndrome, the presence of additional features such as hypertension, easy bruising, and striae help to make the diagnosis more likely.

Hyperthecosis refers to the histological finding of islands of theca cells within dense ovarian stroma, and is almost certainly a variant of PCOS. Its clinical presentation is indistinguishable from that of PCOS, but it tends to be associated with severe hirsutism, and there is often also cutaneous evidence of significant insulin resistance (acanthosis nigricans).

Bullet list 4 Causes of hirsutism

Another well-recognized cause of hirsutism that may be difficult to distinguish clinically from PCOS is nonclassic (late-onset) congenital adrenal hyperplasia resulting from 21-hydroxylase deficiency (Chapter 13.7.2). Such cases tend to present during adolescence with symptoms of anovulation and androgen excess.

Androgen-secreting ovarian and adrenal tumours are rare. Causes, diagnosis, and management of adrenal tumours are described elsewhere (Chapter 13.7.1). Ovarian tumours may be benign or, less commonly, malignant and are classified as either sex cord-stromal tumours (Sertoli–Leydig cell tumours), or adrenal-like tumours (e.g. virilizing lipoid-cell tumours, adrenal rest tumours).

Investigation and diagnosis of hirsutism

A guide to the investigation of hirsutism is given in Table 3. Mild to moderate long-standing hirsutism in women with regular menses is very likely to be either idiopathic or, much more commonly, associated with polycystic ovaries. Serum testosterone concentrations are usually modestly elevated or within the normal range. It could be argued that no investigations are strictly necessary in this category of patient, but our practice is to measure serum testosterone and perform a pelvic ultrasound scan to determine ovarian morphology, so that a specific diagnosis can be offered to the patient. The principal reason for measuring testosterone in women with hirsutism is to screen for the more serious causes of androgen excess that will require further investigation. It is not measured to diagnose hyperandrogenism since this is already clinically manifested as hirsutism. In our clinic we have not found it necessary routinely to measure androstenedione, SHBG, or free testosterone. Some laboratories offer androstenedione as a reasonable alternative to testosterone assays.

In women with hirsutism and menstrual disturbance the most likely diagnosis is again PCOS. In such cases, however, it is legitimate to extend biochemical tests to include measurements of gonadotropins and, in amenorrhoeic women, prolactin and oestradiol. Further investigations are necessary in those patients with a short history of hirsutism (particularly if this is severe), those with symptoms suggesting other endocrine disorders (e.g. Cushing’s syndrome), and those with a serum testosterone that exceeds 5 nmol/litre (normal range 0.5–3.0 nmol/litre). In patients with a serum testosterone concentration in the normal male range (>10 nmol/litre) the presence of an androgen-secreting tumour must be excluded. Imaging of the ovaries and adrenals by MRI is important. In experienced hands, ultrasonography of the adrenals and particularly the ovaries may also be helpful. Selective catheterization of adrenal or ovarian veins to localize a suspected tumour is difficult to execute and rarely informative. Measurement of DHEAS is a useful specific index of adrenal function in patients with a suspected androgen-secreting tumour, but it is less helpful as a routine test in hirsute patients.

The prevalence of nonclassic CAH in our clinic population is very low (<1% of cases of hirsutism), so we do not routinely measure basal and ACTH-stimulated concentrations of 17-hydroxyprogesterone to screen for this disorder. However, it is appropriate to do so in women with severe hirsutism and a serum testosterone greater than 5 nmol/litre.

Management of hirsutism

Further reading