DIAGNOSIS OF HYPERANDROGENISM

DIAGNOSIS OF HYPERANDROGENISM

In its most general sense hyperandrogenism is any clinical or laboratory evidence of androgen excess in women. The most common clinical presentation of hyperandrogenism in reproductive-aged women is hirsutism or acne with or without evidence of anovulation such as oligo- or amenorrhea or dysfunctional uterine bleeding. Elevated blood levels of androgens without clinical symptoms is referred to as cryptic hyperandrogenism.

Hirsutism refers to the presence of course terminal hairs in androgen-dependent areas on the face and body in women. It is differentiated from hypertrichosis, which is excessive growth of thin vellus hair at any body site. Hypertrichosis is usually familial or associated with endocrine disturbances such as anorexia nervosa or thyroid dysfunction, or with medications such as phenytoin, minoxidil or cyclosporin (1). Hirsutism affects between 2-10% of women between the ages of 18 and 45. It is often a source of psychological discomfort and may be a sign of a significant medical disorder as will be discussed. Hirsutism develops when follicles in androgen sensitive areas start to form thick, pigmented (terminal) hair as opposed to thin, short, non-pigmented (vellus) hair normally seen in those areas in women. There have been several scoring systems developed to quantitate hirsutism in order to help standardize the diagnosis and evaluate treatment options. The best known is the Ferriman-Gallwey (F-G) score (2). A popular adaptation of the F-G evaluates nine, instead of the original eleven, body sites (3). Each body site is scored from 0-4 based on the amount of terminal hair present. A score ≥ 8 is consistent with the diagnosis of hirsutism (2), (4). This scoring system has been used most often in treatment studies, but its usefulness is limited because of its inherent subjectivity.

The causes of hyperandrogenism in reproductive aged women can be divided into five categories in descending order of prevalence. The first is polycystic ovary syndrome (PCOS). The second is idiopathic hirsutism. The third category is either adrenal or ovarian steroidogenic enzyme deficiencies. The fourth category is that of ovarian and adrenal androgen secreting tumors. The fifth category is other endocrine disorders such as Cushing syndrome, disorders of cortisol metabolism, hyperprolactinemia and acromegaly. The great majority of women with hirsutism or other symptoms of hyperandrogenism will have either PCOS or idiopathic hirsutism. The remaining causes combined account for less than 5% of the cases of hyperandrogenism (5).

Polycystic Ovary Syndrome

PCOS will be discussed in detail below. The most widely accepted definition of PCOS is that of hyperandrogenism and anovulation with other specific causes such as late-onset 21-hydroxylase deficiency specifically excluded (6,7). PCOS is the most common cause of hirsutism and the most common endocrinopathy in reproductive aged women. It has a prevalence of about 5% in Caucasian and African Americans and in European populations (8-10).

A rare variant Of PCOS is ovarian hyperthecosis. Unlike PCOS, hyperthecosis presents in both pre- and postmenopausal women (11). Women with hyperthecosis often have signs of virilization and have higher testosterone levels (often > 200 ng/dl) and lower LH levels than classic PCOS (6). They are obese and usually have evidence of extreme insulin resistance (12). Ovarian ultrasound demonstrates a decreased number of follicles and markedly increased stroma. On histopathology there are abundant islands of luteinized theca cells throughout the stroma. Women with hyperthecosis are resistance to clomiphene citrate and ovarian wedge resection for ovulation induction. Treatment with gonadotropin-releasing hormone agonist (GnRHa) improves hirsutism (11) and may improve ovarian response to exogenous gonadotropins for ovulation induction (13). Metformin treatment may also increase the likelihood of ovulation in women with hyperthecosis (14).

Idiopathic Hirsutism

Idiopathic hirsutism is excess terminal hair production in androgen dependent areas in the presence of regular ovulation and normal androgen levels (15-18). It is the second most common cause of hirsutism after PCOS and occurs in about 15% of hirsute women. The pathophysiology of this disorder still needs to be fully elucidated, but is thought to be secondary to increased 5-reductase activity in the skin or its appendages, to other alterations in androgen metabolism or to increased sensitivity of the androgen receptor (15,16,19).

Adrenal and Ovarian Steroidogenic Enzyme Deficiencies

Adrenal or ovarian steroidogenic enzyme deficiencies are the most common cause of hyperandrogenism in post-menarcheal women after PCOS and idiopathic hirsutism. Nevertheless these conditions are uncommon to very rare. Late-onset 21-hydroxylase deficiency occurs in 1-5% of hirsute women, with the greatest prevalence in women of Askenazi Jewish descent (20,21).

It has been reported that late-onset 21-hydroxylase deficiency can be reliably excluded by a random 17-hydroxyprogesterone level less than 200 ng/dl (22). Since 17-hydroxyprogesterone is secreted by the corpus luteum, it should not be measured in the luteal phase in ovulatory, hirsute women. A normal baseline measurement of 17-hyroxyprogesteone may not reliably exclude late-onset 21-hydroxylase deficiency in Askenazi Jews. In these patients and in other hirsute women with a random 17-hydroxyprogesterone level greater than 200 ng/dl, an ACTH stimulation test should be performed. A 17-hydroxyprogesterone level greater than 1000 ng/dl one hour after 250 mcg of synthetic ACTH given intravenously establishes the diagnosis of 21-hydroxylase deficiency (20).

Although adrenal and ovarian 3-hydroxysteroid dehydrogenase deficiency has been reported to occur in as many as 10-40% of hirsute women based on modest elevations in 5-steroids in response to ACTH (23), (24), a more strict evaluation of ACTH stimulation results (25,26) and combined adrenal and ovarian stimulation testing (27) suggest that it is very rare. Molecular genetic studies have also failed to confirm mutations of the type II 3 -hydroxysteroid dehydrogenase gene in subjects with mild elevations of 5-steroids by ACTH testing (26,28,29). Convincing cases of late-onset adrenal and ovarian 3 -hydroxysteroid dehydrogenase deficiency are limited to case reports (26,30,31).

11-hydroxylase deficiency causes 5-8% of classic congenital adrenal hyperplasia. However a late-onset form has never been convincingly established as a cause of post-menarcheal hyperandrogenism (32).

There are three case reports of biochemically diagnosed ovarian 17-hydroxysteroid dehydrogenase deficiency (33,34) presenting with PCOS and ovarian hyperandrogenism. However, mutations in the 17-hydroxysteroid dehydrogenase 3 gene, which is responsible for testosterone biosynthesis in the testes and causes male pseudohermaphroditism, are reported to be asymptomatic in females (35). Six girls have been reported with aromatase deficiency. They presented with pseudohermaphroditism, primary amenorrhea, polycystic ovaries and progressive virilization at puberty (36). Thus ovarian enzyme deficiencies as a group are a very rare cause of hyperandrogenism.

Ovarian and Adrenal Tumors

Both adrenal adenoma and carcinoma may present with virilization and hyperandrogenemia (37). Androgen secreting ovarian tumors include Sertoli-Leydig cell tumors, Leydig cell tumors, lipoid or lipid cell tumors and granulose-theca cell tumors (6). Although much discussed, ovarian and adrenal tumors are rare causes of hyperandrogenism. Only one ovarian and one adrenal tumor were found in a series of 350 consecutive hyperandrogenic women (5), and only one ovarian tumor in another series of 478 consecutive hyperandrogenic women (38). Typically women with androgen secreting tumors have abrupt onset of symptoms distinct from menarche and a more rapid progressions of symptoms compared to PCOS. Signs of virilization such as clitoromegaly, frontal balding and deepening of the voice are also more common. Testosterone levels are usually greater than 200 ng/dl or 2 1/2 times the upper limit of normal, but there is clearly overlap between testosterone levels found in tumors and those seen in severe cases of PCOS or hyperthecosis, and the majority of women with high testosterone levels will not have tumors (38). However, the combination of onset distinct from menarche, rapid progression of symptoms, signs of virilization and elevated testosterone levels usually make the diagnosis of tumor straight-forward.

If a tumor is suspected, both ovarian ultrasound and adrenal CT scan should be done to localize it (6). A dehydroepiandrosterone sulfate (DHAS) level less than 700-800 mg/dl does not rule out an adrenal tumor. Only about half of patients with virilizing adrenal tumors will have such a marked elevation of DHAS (37). Although bilateral ovarian and adrenal vein catheterization can determine the source of androgen excess (39), it should probably be limited to premenopausal women wishing to preserve fertility in whom tumor can not be localized by CT or ultrasound. In postmenopausal women with a negative CT scan, a bilateral salpingo-oophorectomy is appropriate for presumed ovarian tumor, as CT is very reliable in detecting even small adrenal tumors.

Other Endocrine Disorders

Endocrine disorders other than PCOS or late-onset 21-hydroxylase deficiency are rare causes of hyperandrogenism (5). Hirsutism may be present in hyperprolactinemia with or without pituitary adenoma, Cushing syndrome and acromegaly. However it is usually not the primary complaint in these disorders. Prolactin should be determined in all patients with anovulation. Cushing syndrome can be ruled out by a normal 24 hour urinary cortisol or normal overnight dexamethasone suppression test (40). If there is any suspicion of acromegaly, a somatomedin-C level (IGF-I) and/or growth hormone suppression test should be done.

POLYCYSTIC OVARY SYNDROME

As noted above PCOS is the most common endocrinopathy in women and is the most common cause of androgen excess, affecting about 5% of reproductive aged women. Although androgen excess in women has been recognized since the time of Hippocrates and had been described in association with diabetes in the nineteenth century, modern investigation into the pathophysiology and treatment of PCOS began with the report of Stein and Leventhal in 1935 that ovarian wedge resection often restored ovulation and fertility to women with hirsutism, obesity, anovulation and polycystic ovaries (41). From that time there continues to be tremendous interest in the clinical and endocrine manifestations, pathophysiology and treatment of PCOS.

Manifestations of Polycystic Ovary Syndrome

The PCOS ovary has a thickened tunica and about twice the cross sectional area of normal ovaries. The increased area is a result of both an increased stromal area and increased numbers of follicles of all stages past the primordial follicle. PCOS ovaries contain 2-3 times the number of 2 mm or greater follicles than do normal ovaries (42). These histologic findings have been confirmed by transvaginal ovarian ultrasound which demonstrates over twice the number of 2-9 mm follicles in PCOS compare to ovulatory controls (43). Polycystic ovaries, as defined on ultrasound by 10 or more 2-8 mm follicles and an increased, echodense stromal area, occur in 70-80% of women who meet the standard diagnosis of anovulation and hyperandrogenism (5,44).

The majority of women with PCOS have ovarian hyperandrogenism as determined by elevated levels of total or free testosterone, which do not suppress normally with dexamethasone, or by abnormal 17-hydroxyprogesterone responses to pituitary-ovarian stimulation with a GnRH agonist. About one half of PCOS subjects have increased adrenal androgen production as defined by increased 17-ketosteroid response to ACTH, with the adrenal being the primary source of androgen excess in about one half of those with adrenal hyperandrogenism (45).

Hypersecretion of LH occurs in about 90% of PCOS subjects as evidenced by an exaggerated LH response to GnRH, elevated mean LH levels, increased LH pulse amplitude and increased LH pulse frequency (46-48). A reversal of the normal pubertal diurnal pattern of LH secretion has been demonstrated in postmenarchal teenagers with PCOS, who have a morning rise in LH that is distinctly different from the nocturnal rise characteristic of normal premenarchal pubertal girls (49).

Women with PCOS are hyperinsulinemic and insulin resistant compared to age and weight-matched controls. Insulin-stimulated glucose utilization is decreased 35-40% in women with PCOS, independently of obesity, a decrease similar to that seen in type 2 diabetes (50). Pancreatic -cell dysfunction also occurs and is more prominent in PCOS women with first degree relatives with type 2 diabetes (51). Consequently about one-third of obese women with PCOS have impaired glucose tolerance and about 10% have type 2 diabetes (52,53). A similar prevalence of abnormal glucose tolerance is found in adolescents with PCOS (54). Women with PCOS have a 5- to 10-fold increased risk for type 2 diabetes compared with age and weight-matched controls (55).

Women with PCOS appear to be at risk not only for insulin resistance and type 2 diabetes but for the other metabolic abnormalities characteristic of syndrome X or the metabolic syndrome including obesity, dyslipedemia, and hypertension (56). Because these abnormalities are tightly associated with increased risk for cardiovascular disease, it is often assumed that the risk is increased in PCOS, however it has yet to be conclusively demonstrated.

About 50% of PCOS women are obese, and obese PCOS have increased abdominal and visceral fat compared to weight-matched controls (57). Lean PCOS subjects have a greater amount of body fat and are more likely to have an android body fat distribution than weight and aged matched controls (58). Both increasing BMI and increasing waist to hip ratio have been associated with increased risk of death from cardiovascular disease and overall risk of death in large, prospective studies (59,60).

Lipid abnormalities in PCOS include lower HDL- and HDL2-cholesterol, higher total and LDL-cholesterol, and higher triglyceride and VLDL-cholesterol levels (56,61). Lipoproteins are similarly affected, apolipoprotein A1 is lower and apolipoprotein B is increased in PCOS. Lipid abnormalities are present in nonobese as well as obese PCOS when compared to weight-matched controls. The differences between PCOS and normal women may be more pronounced at younger ages, which may put PCOS subjects at risk for earlier development of atherosclerosis (62).

There is evidence that elevated plasma levels of hemostatic factors are independently associated with increased risk for coronary heart disease (63). Both abnormal and normal levels of clotting and fibrinolytic factors have been reported in PCOS. Compared to age- and weight-matched controls, PCOS subjects have normal levels or activity of fibrinogen, factor VII, von Willebrand factor, fibrin D-dimer, plasminogen and plasminogen activator inhibitor-1 (PAI-1) (63-65). In addition PCOS patients are not proportionately resistant to activated protein C (66) and have normal homocysteine levels (67), which correlate positively to the degree of insulin resistance (68). However global fibrinolytic capacity is decreased in PCOS compared to age and weight-matched normal women (69). Tissue plasminogen activator (t-PA) antigen is elevated in PCOS. T-PA antigen largely measures t-PA/PAI-1 complexes and correlates independently with coronary heart disease event rate (63). Troglitazone has been reported to significantly decrease PAI-1 levels in PCOS (70).

Endothelial dysfunction is a very early and apparently necessary condition for the development of atherosclerotic plaques. It is found in hypercholesterolemia, obesity, insulin resistance and diabetes (71). There is conflicting evidence for vascular endothelial dysfunction in women with PCOS. Ultrasound measurements of internal carotid hemodynamics are similar or improved in PCOS compared to age-matched controls (72), as is brachial artery flow and reactivity (73). However, endothelin-1, a marker of endothelial injury, is increased in obese and nonobese PCOS compared to age-matched controls, and levels significantly improve after three months of metformin therapy (74). The response of the femoral artery to the endothelium-dependent vasodilators insulin and methacholine is decreased in PCOS compared to age- and weight-matched controls. There was a significant correlation between methacholine response and total and free testosterone and insulin sensitivity (75). These abnormalities were completely reversed by three months of troglitazone therapy (71).

Given the above risk factors, are women with PCOS at increased risk for cardiovascular disease? Increased atherosclerosis at cardiac catheterization has been found in women with a history of hirsutism or with ultrasound confirmed polycystic ovaries (56,76). The compliance of the common and internal carotid arteries is decreased in young women with PCOS (77). PCOS subjects 45 years or older have an increased prevalence of carotid plaques and greater carotid intima-media wall thickness as determined by ultrasound than do age-matched controls (78). The above studies are all consistent with an increased risk of atherosclerosis in women with POCS.

The largest study of long-term consequences of PCOS reported to date is by Wild and Pierpoint and colleagues. They determined risk factors and cause of death in a cohort of 786 PCOS women diagnosed between 1930 and 1979 in the United Kingdom (79,80). Standardized mortality ratios (SMR) were used to compare the causes of death in the PCOS cohort to national rates. PCOS women had a higher prevalence of risk factors for cardiovascular disease including diabetes, hypertension and high cholesterol. The PCOS subjects had a significantly increased risk of death from diabetes (SMR 4.6), but not of death from all causes (SMR 0.93) or cardiovascular disease (SMR 0.78). There was an increased risk of endometrial cancer (odds ratio 6.1) and cerebrovascular disease (odds ratio 3.4) in the PCOS subjects, and a nonsignificantly increased risk of coronary heart disease (odds ratio 1.2). A report from the Nurses' Health Study of the risk of cardiovascular disease in women with irregular menses found a significantly increased risk of coronary heart disease in women who had a history of irregular and very irregular menstrual cycles at ages 20-35 (81). Since the majority of women with oligomenorrhea have POCS, the authors argued that it was a reasonable marker for POCS. The relative risk was 1.53 in women with very irregular periods, which is within the confidence intervals of the risk reported for coronary heart disease in the study by Wild et al. The association was stronger in overweight (BMI >/ 25) than in lean women. The risk was unchanged when women with menopause before the age of 40 were excluded from the analysis.

Pathophysiology of Polycystic Ovary Syndrome

PCOS is a complex, heterogeneous disorder. It is likely genetic, environmental factors contribute to its pathophysiology, and that no single gene mutation will be found that is both necessary and sufficient to cause PCOS. The familial clustering that occurs in PCOS is consistent with a genetic susceptibility. About 50% of sisters of PCOS probands have hyperandrogenemia with or without anovulation, which suggests an autosomal dominant inheritance for a factor predisposing to ovarian hyperandrogenism (82). There have been two recent excellent reviews of the genetics of PCOS (83), (84). Genes most strongly associated with or linked to PCOS include CYP11 which encodes the cholesterol side-chain cleavage enzyme and a region two megabases centromeric from the insulin receptor gene.

Hyperandrogenism is the sine qua non of PCOS. In vitro studies using minced ovarian tissue, adrenal and ovarian vein catheterization studies, both adrenal and ovarian suppression studies, and ovarian stimulation studies with GnRHa (49,85,86) have confirmed that the ovary is the primary source of hyperandrogenism in PCOS. The ovarian hyperandrogenism is a result of increased activity throughout the thecal cell steroid production pathway (87). This increased activity of thecal cell steroid production is intrinsic to the thecal cell because it persists after multiple passages of thecal cell cultures in vitro (88). The genetic basis of this hyperactivity has yet to be identified, but its presence strongly suggests that a primary thecal cell defect can initiate PCOS. A genetic abnormality of steroidogenesis could affect the adrenal as well as the ovary and explain the increased 17-ketosteroid response to ACTH which often occurs in PCOS (89).

The ovarian hyperandrogenism of PCOS is gonadotropin dependent, and gonadotropin suppression with sex steroid or GnRHa results in normal androgen levels (49). It has been reported that 75% of women with clinical evidence of PCOS have an elevated LH level and 94% have an increased LH/FSH ratio (47). These gonadotropin secretory abnormalities have been thought to play an important role in the development of the ovarian hyperandrogenism characteristic of PCOS (90). However, most evidence suggests that the increased levels of LH often seen in PCOS are a result and not a cause of the ovarian steroidogenic abnormality. Thecal cell androgen secretion in response to increasing levels of LH appears to be strictly limited by intraovarian factors. Ovarian hyperandrogenism is more likely a result of escape from down-regulation by these intraovarian factors than a result of elevated LH levels (91). Women with FSH deficiency have no clinical or laboratory evidence of ovarian hyperandrogenism or ultrasound evidence of stromal hyperplasia despite a mean LH level and LH pulse characteristics typical for PCOS (92). Likewise a 40 year old woman with ectopic LH secretion from a benign pancreatic tumor had normal levels of testosterone and androstenedione and normal ovarian morphology by ultrasound (93). Constitutively activating mutations of the LH receptor, which cause gonadotropin-independent precocious puberty in males, do not cause ovarian hyperandrogenism in women (94). Thus, increased LH stimulation alone is insufficient to induce the ovarian stromal hyperplasia and hyperandrogenism characteristic of PCOS.

The LH secretory abnormalities in PCOS are more likely a result of anovulation and ovarian hyperandrogenism than its cause. Progesterone inhibits pulsatile LH secretion, and there is no cyclic production of progesterone by a corpus luteum in PCOS. In addition the pituitary and hypothalamus are less sensitive to the inhibitory effect of exogenous progesterone on LH secretion in PCOS. This progesterone insensitivity can be reversed by an androgen receptor antagonist (95) (19). Thus the LH secretory abnormalities in PCOS appear to be a result of lack of cyclic progesterone production and androgen-induced resistance to the inhibitory effect of progesterone on LH secretion.

An intrinsic ovarian hyperandrogenism may be modified by endocrine factors other than LH. Insulin resistance and hyperinsulinemia occur in approximately 50-70% of PCOS subjects compared to weight-matched controls (55) (20). About 25% of reproductive-aged women with type 2 diabetes have PCOS (96). Women with extreme insulin resistance due to insulin receptor antibodies or to genetic abnormalities of the insulin receptor or post-receptor pathways have severe ovarian hyperandrogenism and virilization (97). PCOS has also been reported in a young woman with an insulinoma and hypoglycemia. Removal of the insulinoma resulted in resolution of the clinical and biochemical features of PCOS (98).

Although a moderate degree of insulin resistance and hyperinsulinemia is neither necessary nor sufficient to cause PCOS, it plays an important role in the pathogenesis of ovarian hyperandrogenism in many cases of PCOS. Both in vitro and in vivo evidence suggests that hyperinsulinemia contributes to excessive ovarian androgen production in PCOS. Insulin synergizes with LH to stimulate thecal cell steroid production to a much greater degree in PCOS than in normal ovaries (97,99), and it inhibits liver production of sex hormone-binding globulin which may increase testosterone bioavailability. Insulin acts via its own receptor in PCOS thecal cells, probably by a signaling pathway separate from that mediating the metabolic effects of insulin (99). Steroids of thecal origin are lowered by drugs that lower insulin levels in PCOS such as diazoxide and insulin sensitizing agents (86,97,100).

The etiology of insulin resistance in PCOS is unclear. An abnormality of insulin receptor autophosphorylation, which persists in long-term culture of fibroblasts, is present in about 50% of PCOS subjects (97). This abnormality is characterized by increased serine phosphorylation, which inhibits the intrinsic tyrosine kinase activity of the insulin receptor. The increased serine phosphorylation of the insulin receptor is appears to be a result of a serine kinase extrinsic to the receptor (101). The insulin resistance in fibroblasts from PCOS subjects is selective, inhibiting the metabolic but not mitogenic pathways of insulin signaling (50).

Skeletal muscle is responsible for about 85% of insulin-stimulated glucose uptake. There is a significant decrease in insulin-stimulated activation of insulin receptor substrate-1-associated phosphatidylinositol 3-kinase activity in skeletal muscle biopsies from obese PCOS subjects, which occurs in parallel with decreased insulin-stimulated glucose uptake in vivo. The decreased kinase activity is not due to decreased amounts of the insulin receptor, insulin receptor substrates or other signaling proteins (102). These findings are consistent with a postreceptor abnormality of insulin signaling similar to that seen in culture fibroblasts from PCOS subjects. However, unlike the findings in fibroblasts, the defect does not persist in cultured muscle cells of PCOS women (50). This suggests that environmental factors may play the major role in the pathogenesis of insulin resistance in PCOS.

Androgen excess may be the environmental factor, which initiates insulin resistance in PCOS. The hyperandrogenism of PCOS favors a central/visceral pattern of body fat distribution, although other hormonal and genetic factors may also play important roles (57). Visceral fat has increased lipolytic activity and free fatty acid levels are increased in PCOS (103). In turn free fatty acids cause insulin resistance in skeletal muscle (50). Thus hyperandrogenemia may favor visceral body fat distribution, which results in relative increases of free fatty acids, which induce skeletal muscle insulin resistance. Insulin resistance similar to the magnitude found in PCOS has been reported in women with nonclassic CAH (104), supporting the hypothesis that insulin resistance can be a secondary effect of mild androgen excess.

In addition to the ability of increased serine phosphorylation to reduce insulin receptor activity, increased serine phosphorylation of cytochrome P450c17, a microsomal enzyme expressed in the ovary and adrenal, increases its 17,20-lyase activity (105). An increase in the lyase activity of this enzyme could result in increased androgen production of both the ovary and adrenal in PCOS. Thus a gain of function mutation of a serine kinase targeting the insulin receptor and P450c17 could result in both the insulin resistance and hyperandrogenism of PCOS. However, when P450c17 is expressed in fibroblasts from PCOS subjects known to have increased insulin receptor serine phosphorylation and in fibroblasts from normal women, no difference is seen in the lyase activity of the enzyme between the two groups, which argues against a single serine kinase affecting both the insulin receptor and P450c17 in PCOS (106).

Insulin resistance in some adults may be a result of reduced fetal growth, which permanently affects postnatal endocrine and metabolic functions. Reduced fetal growth increases the risk of type 2 diabetes, hypertension and coronary heart disease in men and women (107). There is an association of reduced fetal growth with postmenarchal development of hyperinsulinemia, anovulation and functional ovarian hyperandrogenism in girls with a history of exaggerated adrenarche and premature pubarche (108)- (110). Thus antenatal 'programming' may increase the likelihood of postpubertal insulin resistance, ovarian hyperandrogenism and PCOS. It is unknown if hyperinsulinemia of this degree can result in PCOS in the absence of an intrinsic thecal cell abnormality.

The etiology of anovulation in PCOS is often explained by high intraovarian androgen levels which induce atresia and prevent the emergence of a dominant follicle (55,86). However the well known clinical fact that the PCOS ovary is exquisitely sensitive to exogenous FSH argues against the androgen atresia hypothesis because atretic follicles are not FSH responsive. Based on follicular fluid estradiol levels, the proportion of healthy follicles is similar in normal and PCOS ovaries (111).

Androgens increase the number of small, healthy, potentially recruitable follicles in primates. There are 2-3 times the number of 2 mm or greater follicles in the PCOS ovary than in the normal ovary (42). The number of 2-5 mm follicles in the PCOS ovary determined by ultrasound positively correlates with serum testosterone and androstenedione levels (43). In the rhesus monkey, testosterone implants increase the number of healthy pre-antral and small antral but not pre-ovulatory follicles (112). In these follicles, testosterone significantly increases levels of granulosa cell androgen receptor mRNA, which correlates positively with granulosa cell proliferation and negatively with apoptosis (113) (Weil JCEM 7/98). Testosterone also increases granulosa cell expression of FSH receptor mRNA (114). These finding suggests that increased theca cell androgen production may cause the increased number of small, healthy, potentially recruitable follicles found in the PCOS ovary. If so we must begin to think of androgen-induced follicular growth and not androgen-induced follicular atresia as the cause of anovulation in PCOS.

An increased number of recruitable follicles and an overactive granulosa cell compartment is more likely the cause of anovulation in PCOS than is androgen-induced follicular atresia. Granulosa cells from anovulatory PCO ovaries make significantly more estradiol in response to FSH in vitro than do granulosa cells from normal or ovulatory PCO ovaries, which produce similar amounts (111). This increased estradiol response may be a result of the synergistic effect of FSH and insulin on granulosa cells from anovulatory PCO ovaries (115). LH receptors appear on granulosa cells of smaller sized follicles in PCOS ovaries compared to normal ovaries. LH receptors allow aromatase stimulation by LH as well as FSH in the 5-8 mm follicles common in the PCO ovary (116). The cause of anovulation in PCOS may best be thought of as an androgen-stimulated increase in the number of recruitable follicles, which are more responsive to FSH and LH and produce more estradiol and perhaps inhibin than similar sized follicles in normal ovaries. With too many active follicles there is no periodic decline in negative feedback and no FSH rise and selection of a dominant follicle as occurs in ovulatory women with the demise of the corpus luteum. This hypothesis is further strengthened by the finding that menstrual cycles become more regular in older women with PCOS (117) and that cycle regularity is associated with a significant decrease in follicle count by ultrasound (118).

To summarize the pathogenesis of PCOS, an intrinsic abnormality of thecal cell steroidogenesis resulting in ovarian hyperandrogenism is likely to be the central event in most cases. This steroidogenic abnormality which has been demonstrated in long-term thecal cell cultures is assumed to be present in the adrenal as well as the ovary. At least three factors have been identified which may contribute to the insulin resistance of PCOS. The first is hyperandrogenemia which may favor visceral fat distribution and in turn increase free fatty acid levels and lead to resistance to the metabolic effects of insulin especially in muscle tissue. Secondly an intrinsic resistance to insulin may also contribute to lipolysis in the adipocyte and increased free fatty acid levels. The third factor is restricted fetal growth, which may result in postmenarcheal insulin resistance. The hyperinsulinemia may synergize with both LH and ACTH to further increase both ovarian and adrenal hyperandrogenism (119,120). Increased intraovarian androgen levels increase the cohort of recruitable follicles resulting in polycystic ovaries. Anovulation is a result of the increased follicle cohort and increased sensitivity of the granulosa cell to FSH stimulation resulting in tonic estradiol and inhibin production. The increased sensitivity of the granulosa cell to FSH may be a result both of increased intraovarian androgen production and hyperinsulinemia. The elevated LH levels of PCOS are an epiphenomenon and result from lack of periodic progesterone production and androgen inhibition of the negative feedback effect of progesterone.

Treatment of Polycystic Ovary Syndrome

Prevention of Endometrial Cancer

Hyperandrogenic anovulatory women have normal estrogen levels and are constantly exposed to estrogen stimulation of endometrial proliferation. Without the periodic progesterone-induced inhibition of proliferation and differentiation to secretory endometrium that occurs after ovulation, they are at risk to develop endometrial hyperplasia and carcinoma. A threefold increased risk of endometrial cancer in anovulatory women has been reported (121). Increased serum levels of insulin are also associated with an increased risk of endometrial cancer (122). Although 5% or fewer of endometrial cancers occur in women under the age of 40, the majority of such women have PCOS (123).

It is critical that hyperandrogenic women undergo therapy to prevent endometrial hyperplasia or carcinoma. If a patient has been anovulatory for more than a year, an endometrial biopsy is recommended before instituting therapy (124). The oral contraceptive pill (OCP) is an excellent choice, as it both inhibits endometrial proliferation and reduces ovarian androgen production, thus ameliorating the consequences of hyperandrogenism. If a progestin is used alone, studies in postmenopausal women receiving estrogen suggest it should be give for 12-14 days every month to minimize the risk of endometrial hyperplasia (125,126). Insulin-sensitizing drugs may also decrease the risk of endometrial cancer in PCOS by lowering insulin levels and increasing the frequency of ovulation (122) (Diamanti-Kandarakis, JCEM 5/03). However, insulin-sensitizing agents do not uniformly induce ovulation in all PCOS subjects, and their ability to reduce the risk of endometrial hyperplasia or cancer has yet to be demonstrated.

Treatment of Hirsutism

Medical Therapy

The aim of medical therapy is to suppress androgen production, block androgen receptors or decrease the conversion of testosterone to dihydrotestosterone by inhibition of the enzyme 5-reductase.

Oral Contraceptive Pills - OCPs have commonly been used to treat patients with hirsutism and other signs of androgen excess. The progestational component of the OCP inhibits pituitary secretion of LH, which in turn decreases ovarian androgen production. Progestins also decrease adrenal DHAS production, possibly via a negative feedback loop through the glucocorticoid receptor (4,19). In addition, the estrogen component of oral contraceptive pills increases production of SHBG thus decreasing the amount of free testosterone available (86). All formulations of low dose (≤ 35g ethinyl estradiol) oral contraceptive pills available today, despite the touted differences in their progestational components, have low androgenic potential and are probably equally efficacious in treating acne and hirsutism (127). They are an excellent choice for patients with abnormal cycles and acne but used alone are not the most efficacious therapy for hirsutism (128,129).

Gonadotropin Releasing Hormone Agonists - Administration of a long-acting gonadotropin GnRHa such as leuprolide acetate suppresses ovarian androgen production by inhibiting pituitary gonadotropin secretion. This results in decreased levels of circulating testosterone and androstenedione with no effect on adrenal androgens (130-132). Several investigators have demonstrated clinical improvement of hirsutism especially in women with PCOS (ovarian hyperandrogenism) treated with GnRH agonists (130,133,134).

A prospective, randomized study by Azziz et al. revealed that treatment with leuprolide acetate and cyclic estrogen/progestin therapy resulted in a more complete and rapid improvement in hirsutism by subjective and objective measures compared to OCP use alone (128). This finding was supported by two other groups of investigators who found GnRHa to be superior to OCP use (129,135). A similar randomized prospective study by Carr et al. found no difference at three and six months in objective measures of hair growth in patients treated with GnRHa and OCP, GnRHa alone or OCP alone (136). However, they did find the greatest and most rapid suppression of circulating androgens at three months in the group that received GnRHa and OCP with fewer side effects than GnRH agonist alone.

Because the profound hypoestrogenism caused by GnRH agonists limits the length of treatment time and because hirsutism rapidly reappears after stopping therapy (137), the addition of an OCP may have some advantages. It will prevent bone loss and the vasomotor symptoms associated with hypoestrogenism (136), while still maintaining the advantage of rapid androgen reduction associated with GnRHa use. In addition, the continued use of OCP after GnRHa therapy has been discontinued may prevent the rapid recurrence of hirsutism (137), although this requires further study. Given the cost of GnRHa, they are better reserved for patients who fail first line therapy with an oral contraceptive and antiandrogen.

Glucocorticoids - Hirsutism secondary to late-onset congenital adrenal hyperplasia or functional adrenal hyperandrogenism may respond to treatment with low doses of glucocorticoids, such as 5 mg of prednisone or 0.25 mg of dexamethasone every other day (4,138). However, even women with these underlying conditions respond better to traditional antiandrogen therapy (139,140).

Androgen Receptor Antagonists - Although not available in the United States, cyproterone acetate was the first androgen receptor antagonist to be used clinically and is still widely used in Europe (138,141). It is a competitive inhibitor of testosterone and dihydrotestosterone receptor binding and also has progestational and weak glucocorticoid properties (141,142). It is an effective and well-tolerated treatment for hirsutism.

Spironolactone is a competitive inhibitor of the aldosterone receptor and was initially utilized as a potassium sparing diuretic. It was soon discovered to have antiandrogenic properties, and when used together with the OCP, it is the first line treatment for hirsutism in the United States. Its antiandrogenic effects come from several mechanisms, the most important of which is the blockade of androgen receptors in the hair follicle (143). In addition spironolactone also inhibits androgen biosynthesis through the cytochrome p450 system and directly inhibits 5-reductase activity.

Treatment with spironolactone should begin at a dose of 200 mg/d for at least 3-6 months (144-146). Efficacy of spironolactone alone has been quoted between 75-95% using subjective measures such as patient satisfaction and the Ferriman-Gallwey score, and at-least two studies have documented an objective decrease in hair shaft diameter with use of spironolactone alone (146-148).

Treatment with spironolactone is generally very will tolerated with the most common side effects being irregular vaginal bleeding, polyuria and fatigue (147). It is important to remember that with spironolactone, as with all antiandrogens, pregnancy can still occur with the theoretical potential for feminization of male fetuses. For that reason the OCP is often used in conjunction with spironolactone. Not only will it protect against pregnancy, but also control abnormal uterine bleeding and possibly potentiate the effect of spironolactone (149).

Flutamide is a nonsteroidal antiandrogen that appears to work only at the androgen receptor (150). Flutamide 250 mg/d for six months is effective in treating hirsutism, and in two randomized trials it was found to be equivalent to treatment with 100 mg/d of spironolactone (151-153). Patients with more severe hyperandrogenism or alopecia, may respond better to flutamide than to spironolactone as demonstrated in a randomized controlled trial (154). The most common side effects of flutamide are mild and include dry skin and increased appetite. However, the potential exists for a rare but severe drug-induced hepatitis which limits the usefulness of this medication (138). Because of this potentially severe side effect, it is generally recommended that flutamide be utilized after other therapies have failed and that liver transaminases are monitored appropriately.

The effects of long-term therapy with antiandrogens are largely unknown. Certain breast cancer cell lines exhibit androgen receptors and the action of androgens in vitro is to inhibit breast cancer cell line proliferation. This action can be inhibited by the antiandrogen hydroxyflutamide (155). It is not known at this time what if any effect long term use of antiandrogens would have on breast tissue and more specifically on the development and growth of breast cancer in vivo.

5-reductase Inhibitors - Finasteride is a potent inhibitor of 5-reductase and thus reduces the conversion of testosterone to its active metabolite dihydrotestosterone (156). It was initially used to treat benign prostatic hypertrophy and recently a lower dose (Propecia , 1 mg/d) has been approved in the treatment of male pattern baldness. Two non-randomized studies of finasteride and one randomized placebo controlled study revealed significant decreases in the Ferriman-Gallwey hirsutism score after 3-6 months of treatment at 5 mg/d (157-159). This reduction appears to be potentiated with concomitant use of oral contraceptive pills (159). Given the significant potential of 5-reductase inhibitors to feminize the external genitalia of a male fetus, good birth control must be employed.

Finasteride is well tolerated with minimal side effects at the standard dose of 5 mg/day. A randomized, double blind, placebo controlled study by Moghetti et al. objectively compared the effectiveness of spironolactone, flutamide and finasteride by using computer assisted light microscopy determination of hair shaft diameter in hirsute women. At the end of a six- month period all three-treatments similarly reduced hair diameter and the Ferriman-Gallwey score compared to the placebo group (153).

Insulin Sensitizing Agents - A randomized, double blind, placebo-controlled trial of increasing doses of troglitazone in 305 women with PCOS showed a 15% decrease in the Ferriman-Gallwey score at the highest (600 mg/day) dose of troglitazone after 20 weeks of therapy (160). Metformin has been shown to decrease the Ferriman-Gallwey score and objective hair growth rate by about 15% in women with PCOS in a double blind, placebo-controlled cross over study (161) and to decrease the Ferriman-Gallwey score by 15% in PCOS subjects on a weight loss diet randomized to placebo or metformin in a double blind fashion for 6 months (162). An unblinded trial also found metformin to be effective in reducing the Ferriman-Gallwey score in adolescents with PCOS (163). These studies were on small numbers of patients (n = 10-13) and should be considered preliminary. Overall the effect of insulin sensitizing agents on hirsutism appears to be modest, and we found no studies of these agents compared to standard OCP/antiandrogen therapies.

Eflornithine HCL - Eflornithine HCL (difluiromethyl ornithine) 13.9% is available in a new product for excessive facial hair growth, called Vaniqa (Brisol-Myers Squibb). This compound irreversibly blocks the enzyme ornithine decarboxylase (ODC) which catalyzes the rate limiting step in the synthesis of polyamines (164). ODC is concentrated in the matrix of growing hair follicles and inhibition of polyamine synthesis impairs cell differentiation and division, decreasing hair growth and converting new growth to finer hair (165). The idea of treatment with eflornithine is to decrease the frequency of and improve results of mechanical means of hair removal.

There have been two phase 2, randomized, double blind, placebo-controlled clinical trials involving 594 women with the diagnosis of hirsutism (21). These trials compared treatment with eflornithine 13.9% vs. placebo for 24 months on facial hair only. They found a statistically significant difference at 8 weeks by subjective and objective measures in the treatment group. This improvement continued up to 24 weeks with 32% of the patients improved in the treatment group compared to 8% in the placebo group.

Summary - A reasonable approach to medical therapy for women with hirsutism is to combine the oral contraceptive pills with an antiandrogen. Since the efficacy of all antiandrogens is similar at the doses specified, spironolactone is the first choice in the United States because it is the cheapest and has minimal side effects. If satisfactory results are not obtained with this therapy in six to nine months, one could consider adding a GnRH agonist for more complete suppression of ovarian androgen production or substituting finasteride for spironolactone or adding finasteride to spironolactone. There are no clinical studies of such therapies in women failing initial medical therapy to guide one's decision. Eflornithine may be most appropriate for women with relatively mild facial hirsutism, or it may be a reasonable additional therapy in women with inadequate response to the oral contraceptive pill and spironolactone. Flutamide should be reserved for recalcitrant cases given the rare complication of hepatitis.

Non-Medical Therapy - Medical therapy with OCP and antiandrogens will decrease production of new terminal hair in androgen dependent sites. However, it will not immediately affect hair already present, and it will not affect hair growth at androgen independent sites. Therefore different means of mechanical hair removal have been employed in combination with medical therapy to achieve optimal results.

Shaving - Shaving is a fast, simple and inexpensive means of temporary hair removal. However, hair quickly regrows at a normal rate and the cosmetic results are typically unappealing to women for use on the face. There have been some studies in mice revealing that shaving may actually induce hair to change from telogen to the anagen or growth phase (166).

Epilation - Epilation includes plucking and waxing and involves removal of the hair from the bulb. It does not change the rate or duration of hair growth but repeated plucking may lead to a delay in the return to anagen and thinner hair secondary to permanent matrix damage (167). Epilation is an acceptable means of hair removal. Again it is only temporary although it may last 2-3 weeks longer than shaving. However, it is costly and may be associated with pain and inflammation at the site.

Depilatories - Chemical depilatories are a painless method of hair removal that generally last slightly longer than shaving without bristly regrowth of hair. All chemical depilatories act by breaking disulfide bonds in hair causing it to break off at or just beneath the surface of the skin (168). The most common side effect is an irritative dermatitis. Similar to other products for the treatment of hirsutism, the effects are reversed shortly after discontinuation of therapy.

Electrolysis - Electrolysis is a permanent method of hair removal that utilizes electrical current (direct or alternating) to destroy the hair follicle (169). This is a slow technique that may take months of repeated therapy to treat an area. For this reason, electrolysis is best reserved for small, localized areas of hair growth. Even in experienced hands 15-25% of treated hair will regrow (170). Side effects of electrolysis consist of pain during and shortly after treatment, scarring, pigmentation changes and local infection (167).

Lasers - The use of lasers for the removal of unwanted hair has increased over the past decade due to the increased demand in our society for better alternatives to remove hair. Lasers take advantage of selective photothermolysis in order to selectively destroy the hair follicle. This implies that a specific target (chromophore) is utilized that can absorb light at a specific wavelength. The most common chromophore that lasers are targeted against is melanin, which is highly concentrated in the hair follicle but not the surrounding dermis. This allows for selective destruction of the follicle without damaging nearby structures. The epidermis however, contains melanin and the laser system must be equipped with a method to cool the skin to protect the epidermis from thermal damage (169). The most commonly used lasers are the ruby lasers (694 nm) and the new long-pulsed alexandrite lasers that have a lower risk of epidermal injury secondary to the longer wavelength. Similarly, the powerful 800 nm diode lasers are safer for darker skinned individuals because the epidermis does not absorb the longer wavelength light as easily (171).

Photodynamic therapy utilizes light and a chemical photosensitizer to permanently destroy the hair follicle. There are several agents under investigation and development for use with laser therapy. The best studied is the topical use of aminolevulinic acid (ALA). ALA induces the synthesis of a potent photosensitizer, protoporphyrin IX in the hair follicle, that when activated by red light will cause cell membrane damage by the formation of oxygen free radicals (172). The advantage of photodynamic therapy is the ability to quickly treat larger areas and use on more skin types. However this treatment is still in the developmental stage.

The effectiveness of laser treatment is based on the type used and patient selection. The ideal patient for laser hair removal is fair skinned with dark pigmented hair (171). Lasers induce a temporary complete hair loss meaning that a percentage of the hair will eventually grow back (168). However the regrowth is generally thinner and lighter than the original hair. Approximately 80% of patients respond positively to laser hair removal, although a certain percentage of patients will need multiple treatments (169).

Most patients experience some discomfort during and shortly after the procedure. Perifollicular edema and erythema is common and may last up to 3 days. Epidermal damage leading to bacterial or HSV infection or pigmentary changes is uncommon if performed properly. Darker skinned or tanned individuals are at more of a risk for permanent pigmentary changes and proper precautions should be taken such as preoperative sunscreen or bleaching agents (169). If performed by qualified personnel, laser hair removal offers an excellent alternative for permanent hair reduction.

Treatment of Insulin Resistance and Other Metabolic Abnormalities

Weight Loss

Treatment of the metabolic abnormalities of PCOS has the potential to reduce the risk of developing type 2 diabetes and cardiovascular disease. It is important to emphasize that the best method to increase insulin sensitivity is weight loss. Weight loss is more effective than metformin in reducing the rate of progression to diabetes in subjects at high risk for diabetes (173). Several studies have documented the salutary effect of modest loss of 5 to 10% of body weight in obese PCOS (174-177). An improvement in menstrual function has been reported in as many as 80-90% of patients. Fasting insulin and/or insulin response to glucose tolerance testing decrease significantly and insulin sensitivity as determined by euglycemic insulin clamp improves significantly after weight loss. Total cholesterol, LDL-cholesterol and trigycerides decrease by about 10% (178). The pregnancy rate increases with both timed intercourse and in treatment cycles and the spontaneous abortion rate decreases (176,179). Weight loss also has a consistent effect on hyperandrogenism with significant decreases to normal or near normal levels of total and free testosterone and a significant increase in sex hormone-binding globulin.

Insulin-Sensitizing Agents

Metformin is the insulin-sensitizing agent of choice in PCOS. It is a pregnancy category B drug, and it is the drug used in the great majority of published studies on insulin-sensitizing agents in PCOS. Metformin therapy is begun at 500 mg/day with dinner and is increased by 500 mg increments every 1-2 weeks as gastrointestinal symptoms abate. Most studies in PCOS have used doses of 1500-1700 mg/day given in 2-3 doses. An extended release formulation is also available for a single daily dose. Metformin treatment usually results in modest weight loss, which may be due in part to its suppression of appetite. The most common side effects of metformin are gastrointestinal and include diarrhea, flatulence, nausea and abdominal discomfort. These symptoms usually can be minimized by slowly increasing the dose. However about 5% of patients cannot tolerate metformin because of gastrointestinal side effects. (180). The most serious complication of metformin is lactic acidosis which occurs at a rate of about 3 cases per 100,000 patient years of use. About 90% of metformin-related cases of lactic acidosis had a predisposing condition such as congestive heart failure, renal insufficiency, chronic lung disease with hypoxia or age older than 80 years (180).

Metformin in doses of 1500-1700 mg/day appears to be effective in improving both reproductive and metabolic outcomes in PCOS. Most studies have found that metformin decreases total and/or free testosterone levels (100,162,181-183). Although not completely consistent studies have found that metformin significantly lowers BMI, visceral adipose tissue, fasting insulin and insulin response to glucose (162), (184). Some (182), (185) but not all (183) studies have shown improved insulin sensitivity. Significant increases in HDL-cholesterol have also been reported (182,185). Women with PCOS lose significantly more weight when treated with metformin and weight loss diet than with placebo and weight loss diet (162). When metformin is given to very obese patients who maintained their weight during the 12 weeks of therapy, there is no change in free testosterone, in insulin response to a glucose challenge or in insulin sensitivity, which suggests that the effects of metformin may be due in part to weight loss (186). When given to hyperandrogenic hyperinsulinemic adolescent girls with a history of premature pubarche, metformin decreases insulin response to glucose, decreases total and LDL-cholesterol and triglycerides and increases HDL-cholesterol (163). It also increases lean body mass and reduces abdominal fat in this same population (187).

There is no information in PCOS subjects on whether or not metformin decreases the progression from impaired glucose tolerance to type 2 diabetes. However because the Diabetes Prevention Program found it effective in middle-aged subjects at risk for type 2 diabetes (173), it would seem prudent to perform yearly oral glucose tolerance testing on PCOS patients and treat those with impaired glucose tolerance with metformin and/or an aggressive weight loss program.

Thiazolodinediones, which are pregnancy category C drugs, are another class of insulin-sensitizing agents that have been studied in PCOS. Most studies are with troglitazone, which was withdrawn from the market by the manufacture because of the rare complication of hepatocellular damage leading to acute hepatic failure in patients with type 2 diabetes. Two new thiazolodinediones, rosiglitazone and pioglitazone, were approved for the treatment of type 2 diabetes in 1999 and appear to have an even lower risk than troglitazone of hepatocellular damage. Troglitazone given for 12-44 weeks decreases total and/or free testosterone in a dose-dependent fashion (70,160),188). Insulin sensitivity and insulin response to an oral glucose challenge also improved in a dose-dependent fashion. In the largest study of 305 PCOS subjects randomized to placebo, 150, 300 or 600 mg troglitazone a day, ovulation rate increased in a dose-dependent manner with evidence of ovulation as determined by frequent urinary pregnanediol measurements in 62% of 4 week intervals at the 600 mg dose (160). There is no significant effect of troglitazone on lipids (70), and weight either remains unchanged (70), (188) or increases a small but significant amount (160). Pioglitazone or rosiglitazone have similar favorable effects to troglitazone on ovulation, androgen levels, insulin sensitivity, and fasting and glucose-stimulated insulin levels in small studies of 18-25 women with PCOS (189,190).

Oral Contraceptive Pills and Antiandrogens

The metabolic effects of OCPs and antiandrogens in PCOS are of great interest because both classes of drugs are often given for long periods of time to treat hirsutism or to prevent endometrial hyperplasia. As a group antiandrogens produce modest improvements in insulin resistance and lipid levels after 3-4 months of therapy in small numbers of patients. The antiandrogens spironolactone and flutamide as well as the GnRH agonist, buserelin significantly increase insulin-mediated glucose disposal in hyperandrogenic women, but not to the level of that seen in weight-matched controls (191). Flutamide significantly lowers total and LDL-cholesterol and triglycerides after 12 weeks of therapy in lean and obese PCOS subjects (192). When given to adolescent PCOS subjects for six months, flutamide, 250 mg twice a day resulted in ovulation in all subjects (193).

Recently concern has been expressed that the OCP may unfavorably affect metabolic parameters in PCOS (122). This concern is based in part on reports of deterioration in glucose tolerance in small numbers of POCS subjects treated with the OCP. In obese women with POCS, diabetes developed in two of seven with impaired glucose tolerance, and impaired glucose tolerance developed in five of nine with normal glucose tolerance treated with a desogestrel-containing OCP for six months (194). In a second study of 10 obese PCOS subjects treated with a 2 mg cyproterone acetate-containing OCP for six months, three developed impaired glucose tolerance and one with baseline impaired fasting glycemia developed diabetes (183), despite the fact that there was no effect of the OCP on insulin sensitivity as determined by euglycemic hyperinsulinemic clamp.

OCPs have varying effects on insulin sensitivity as determine by clamp studies. A triphasic norethindrone OCP given for three months resulted in a modest decrease in insulin sensitivity in normal and PCOS subjects (195). Cyproterone acetate 100 mg given in the reverse sequential protocol with 50 ug ethinylestradiol for 6 months also resulted in a modest decrease in insulin sensitivity (196). However 2 mg cyproterone acetate given together with 35 ug ethinyl estradiol in a standard, combination OCP for six months had no effect on insulin sensitivity in obese (183) or nonobese PCOS (197,198). Likewise a norgestinate OCP did not affect insulin sensitivity after six months of therapy in non-obese PCOS (199).

Metformin and the 2 mg cyproterone acetate/35 ug ethinyl estradiol OCP have been compared to the OCP alone in one randomized trial in nonobese PCOS. After four months of therapy the metformin and OCP group had a significant decrease in BMI and WHR and a significant increase in the glucose to insulin ratio, while the parameters were unaffected in the OCP only group (200).

There is one long-term, observational study of the OCP in young women with PCOS followed for a mean of 10 years (201). During that time 16 POCS subjects used the OCP for a mean of 97 months and 21 never used the OCP. OCP users had a significant decrease in WHR and in glucose response to an oral glucose tolerance test. The nonusers had a significant increase in fasting glucose and in insulin response to an oral glucose tolerance test.

The studies of the metabolic effects of antiandrogens and OCPs in PCOS are limited by small sample sizes and short treatment periods. Antiandrogens and GnRHa produce modest benefits on metabolic parameters. The effect of OCPs is more complex. OCPs containing lower doses of progestins or progestins of lower androgenicity may have fewer unfavorable effects than higher dose or higher androgenicity OCPs. However the most important consideration may be patient weight. Glucose tolerance may deteriorate in obese PCOS treated with the OCP. Obese subjects may be better served by periodic progestin-only therapy to induce withdrawal bleeding and reduce the risk of endometrial hyperplasia, or by a combination of the OC and metformin. However the ideal therapy has yet to be determined. It is possible that long-term suppression of androgens by the OCP may improve or prevent deterioration of the metabolic profile of younger POCS subjects.

Treatment of Anovulation

Anovulation is the primary cause of infertility in about 20% of couples, and PCOS is estimated to be the cause of 70% of anovulatory fertility (181,202). There are many therapies for the induction of ovulation in PCOS patients. The general paradigm is to begin with the easiest to manage therapies, and if these do not result in ovulation or pregnancy in a reasonable period of time, to move on to more elaborate therapies.

Clomiphene Citrate

Clomiphene citrate is still the first line of therapy for ovulation induction in women with PCOS (138), (203), although the argument has been made that metformin is preferable (204). As of yet there are no large randomized clinical trials comparing the two as initial therapy for ovulation induction. The standard clomiphene regimen is 50 mg /day for 5 days beginning on cycle day 3-5 following spontaneous or progestin-induced bleeding. If serum progesterone in the mid luteal phase is less than 10 ng/mL, the dose can be increased by 50 mg a day in subsequent cycles to a dose of 150 mg/day.

Although there is little information about ovulation induction in PCOS specifically, large retrospective studies have shown that about 90% of women with oligomenorrhea ovulate following standard clomiphene citrate therapy, and of those about half conceive (205,206). There is a 15% spontaneous abortion rate and a 4% incidence of twins. In women without other infertility factors, about 88% of those ovulating will eventually conceive with a monthly fecundity rate of 0.22, which is similar to that of fertile women discontinuing the diaphragm. Factors which favor ovulation and pregnancy in clomiphene-treated patients include oligomenorrhea (versus amenorrhea), lower free androgen index, lower BMI and younger age (207). If clomiphene citrate successfully induces ovulation, the primary cause of failure is premature termination of therapy. In patients with no other infertility factors, at least six clomiphene citrate-induced ovulatory cycles are appropriate before moving on to other therapies.

About 20-25% of women with PCOS fail to respond to standard clomiphene citrate regimens (202), (207). In these patients some success has been reported using doses up to 250 mg a day for 5-22 days. These high dose, extended protocols may increase the rate of ovulation, but they are cumbersome to use and pregnancy rates are only about 12% per ovulatory cycle (138).

Metformin

Metformin in doses of 1500-1700 mg/day significantly increases rates of spontaneous ovulation and clomiphene-induced ovulation compared to placebo or placebo and clomiphene (181,185,208). Spontaneous ovulation rates of 34-82% have been reported on metformin alone, and ovulation rates of 80-90% have been reported with metformin and clomiphene (50-150 mg/day for 5 days). Metformin is also effective in inducing ovulation in adolescent girls with PCOS and a history of premature pubarche (209). There have been four randomized, blinded studies of metformin and clomiphene in PCOS who failed to ovulate with clomiphene (100-150 mg/day for 5 days) alone (210-213). Two studies found metformin and clomiphene to be superior to clomiphene alone and two found no statistically significant difference. All four trials were relatively small (19-56 total subjects). Metformin was effective in the largest trial, and in one trial where it was not, the mean BMI of the study subjects was low compared to the other three trials (24 vs. 31-38). These studies suggest a role for metformin in ovulation induction, but a definitive, large, multicenter trial has yet to be reported.

Two randomized trials have used metformin together with FSH for ovulation induction (214,215), and there is one study of ovarian stimulation with FSH and metformin for in vitro fertilization (216). Again the total number of patients was small, and the IVF study was retrospective. Metformin appears to decrease the number of small follicles, lower serum estradiol and decrease the risk of ovarian hyperstimulation. In the patients undergoing IVF, there were significantly more mature oocytes collected and a significantly higher pregnancy rate (70 vs. 30%) in the metformin treated patients.

Dexamethasone

Another alternative to standard clomiphene citrate therapy is to treat with a short course of dexamethasone 0.5-2 mg daily together with clomiphene citrate. Two randomized trials have demonstrated significantly improved rates of both ovulation (about 90% versus 20-35%) and pregnancy (40-75% versus 5-35%) in patients treated with clomiphene and dexamethasone versus clomiphene alone (217,218).

The mechanism of dexamethasone treatment is unknown. It has been hypothesized that it reduces adrenal androgen production thereby reducing peripherally produced estrogens and estrogen negative feedback on FSH secretion or that reduced adrenal androgens may also result in decreased peripheral conversion of DHAS and androstenedione to testosterone and thereby reduce follicular atresia. It is intriguing that both dexamethasone and insulin sensitizing agents improve the rate of ovulation in response to clomiphene despite their disparate effects on insulin sensitivity. Dexamethasone increases insulin resistance and may induce diabetes in at-risk individuals (219). If dexamethasone increases insulin resistance and hyperinsulinemia, why does it improve the rate of ovulation in response to clomiphene? One hypothesis would be that the mechanism by which dexamethasone induces insulin resistance does not spare the ovary. That hyperinsulinemia has adverse effects on ovarian function by increasing both granulosa and thecal cell steroid production implies that the insulin resistance in PCOS spares the ovary. Dexamethasone may ameliorate the steroidogenic effect of insulin in the ovary by inducing ovarian insulin resistance, and thereby paradoxically improving the likelihood of ovulation.

Gonadotropins

Options for women unresponsive to standard or modified clomiphene citrate stimulation therapies or to metformin alone include stimulation with gonadotropins or surgically induced ovulation. Issues regarding these options include whether or not human recombinant FSH (rFSH) or purified urinary FSH (uFSH) is superior to human menopausal gonadotropins (hMG), whether low dose gonadotropin administration is superior to standard doses, whether there is an advantage to pituitary down-regulation with GnRHa , whether there is a role for surgically induced ovulation in contemporary practice, and finally which method best limits the spontaneous abortion rate.

Early studies showed that hMG successfully induced ovulation in 95-100% of clomiphene citrate resistant PCOS (220,221). However, pregnancy rates were significantly lower than in women with hypogonadotropic hypogonadism treated with hMG (91 vs. 36% after 6 cycles of therapy) (222). Spontaneous abortion rates also appear to be relatively high (21-42%) and significantly greater than that seen in women with hypogonadotropic hypogonadism (220,221,223).

Although virtually all physicians experienced in hMG therapy believe ovarian hyperstimulation is more common in PCOS than in hypogonadotropic hypogonadism, this clinical impression was not confirmed by early, large studies that showed hyperstimulation rates of 4% based on clinical and urinary estrogen criteria (221,223), which were not significantly different from the rate in hypogonadotropic hypogonadism. Likewise multiple gestation rates of 16 - 36% were reported, and Oelsner et al. found the multiple gestation rate significantly higher in hypogonadotropic hypogonadism than in PCOS (38 vs. 16%).

In attempts to decrease the perceived high rates of ovarian hyperstimulation and multiple gestation in PCOS with standard gonadotropin therapy, several groups have reported on low-dose gonadotropin stimulation. With step-down therapy 150 IU of gonadotropin is given daily until the largest follicle is at least 10 mm in diameter. At that point the dose is decreased by 37.5 IU. If follicular growth continues, the dose is decreased to 75 IU three days later and maintained at that dose until hCG is given when at least one follicle is 18 mm in diameter (224). Most reports utilize step-up therapy where stimulation is initiated with 50-75 IU/day of gonadotropins, and patients are maintained on this dose for 10 days to two weeks. If no follicle develops to at least 10 mm by 14 days of therapy, the dose is increased by 37.5 IU every seven days as necessary. Compared to standard gonadotropin therapy, low dose therapy with either uFSH or hMG results in a higher rate of single dominant follicle development, fewer ovulatory follicles at the time of hCG administration (usually defined as follicles greater than or equal to 15 mm in diameter) and lower mean estradiol level (225-229). A recent randomized, multicenter trial found the step-up protocol to result in higher rates of monofollicular development and hCG administration than the step-down protocol (230).

Hyperstimulation based on ultrasound and clinical criteria was also less likely with a low dose protocol (228). In a report of 225 PCOS patients treated with the step-up protocol, rates of ovulation and pregnancy were similar to those of a standard protocol with 95% of patients ovulating and 45% conceiving after a mean of four cycles per patient (231). Multiple gestation occurred in only 6% of pregnancies, and all were twins. However, spontaneous abortion rates remained high, at about 30%.

uFSH or rFSH has a theoretical advantage of avoiding additional LH in PCOS, who often have elevated LH levels. However, uFSH at standard doses (150 IU/day with 75 IU increases every 3-5 days as necessary) gives results similar to hMG for rates of ovulation, pregnancy, hyperstimulation, multiple gestation and spontaneous abortion (232-234). As with standard gonadotropin therapy, there is no clinical advantage of low-dose uFSH over low-dose hMG (235). Likewise there is no advantage of rFSH over uFSH regarding ovulation, pregnancy, miscarriage or hyperstimulation rates (236).

Because gonadotropin therapy in PCOS give disappointing results compared to hypogonadotropic hypogonadal patients, investigators and were quick to investigate GnRHa down-regulation of pituitary-ovarian function in PCOS prior to gonadotropin therapy. It was hoped that pregnancy rates would improve and ovarian hyperstimulation would be reduced by transforming the PCOS patient to a hypogonadotropic hypogonadal state with GnRHa.

Early reports in small number of patients were encouraging with excellent pregnancy rates of 27-54% per cycle, when GnRHa down-regulation was combined with hMG therapy (138). Subsequent prospective studies of low-dose gonadotropin therapy vs. low dose gonadotropin and GnRHa down-regulation found significantly increased risks of multiple follicle development and ovarian hyperstimulation in the GnRHa groups, with no increase in pregnancy rates (237,238). Thus, inhibition of pituitary gonadotropin secretion by GnRHa increases the risk of ovarian hyperstimulation without increasing the pregnancy rate in women with PCOS, but there is some evidence that GnRHa may decrease the rate of spontaneous abortion (see below).

Surgical Ovulation Induction

The final option in treating clomiphene citrate resistant PCOS is surgical ovulation induction. Bilateral ovarian wedge resection was the primary therapy to induce ovulation in PCOS for many years. About 80% of patients treated ovulated at some point after therapy and about 55% achieved pregnancy (239). However surgical therapy fell out of favor because of the availability of clomiphene citrate and because of the concern that the high prevalence of adhesions following wedge resection would affect fertility.

More recently there has been great interest in using laparoscopic methods to surgically induce ovulation in women with PCOS. Electrocautery or laser is used to produce multiple burns in the ovarian capsule, or ovarian tissue may be excised with scissors. All laparoscopic methods of ovulation induction result in similar rates of ovulation and pregnancy. A recent review found ovulation rates of 77% and pregnancy rates of 60% in a total of 1124 reported patients (239). The rate of miscarriage was 14% and the multiple pregnancy rate was only 2.5%. Most patients ovulate spontaneously following laparoscopic ovulation induction, but some require clomiphene citrate or gonadotropin therapy.

Some degree of adhesion formation has been reported in 5-71% of patients after laparoscopic surgery (239). Although a case of severe adhesions and bilateral tubal occlusion after ovarian electocautery has been reported (240), adhesions are usually mild, and adhesions severe enough to limit fertility appear to occur infrequently after laparoscopic surgery for ovulation induction in PCOS. It is a theoretical concern that destruction of oocytes during laparoscopic procedures may increase the risk of premature ovarian failure (241).

The advantage of laparoscopic ovulation induction in PCOS is that it may result in long-term ovulation without any other therapy. Miscarriage and multiple pregnancy rates are low, but similar to those seen with low-dose gonadotropin therapy (242). Surgical procedures for ovulation induction tend to fall out of favor as medical therapies improve. Although there are no randomized trials comparing laparoscopic surgery and metformin in clomiphene resistant subjects, reported outcomes are similar (241).

Spontaneous Abortion

In ovulatory women both elevated LH levels and polycystic ovaries on ultrasound are reported to increase the risk of spontaneous abortion (243,244). However, there is controversy as to whether or not this ovulatory, PCO-like syndrome is a cause of recurrent spontaneous abortion. In women with a history of recurrent spontaneous abortion, there is no effect of LH or testosterone levels, BMI or PCO by ultrasound on subsequent risk of spontaneous abortion (245). A randomized trial of pituitary-ovarian down-regulation with GnRHa followed by low dose gonadotropins vs spontaneous ovulation in 106 women with PCO, elevated LH and recurrent spontaneous abortion resulted in 26/40 (65%) live births with treatment and 35/46 (76%) live births with spontaneous ovulation (246).

As noted above anovulatory women with PCOS appear to be at increase risk for spontaneous abortion following ovulation induction. Retrospective studies suggest that pituitary-ovarian down-regulation with GnRHa followed by low dose gonadotropins may decrease the abortion rate in anovulatory women with PCOS (138); however, this has never been verified in a prospective randomized trial.

Women with recurrent spontaneous abortion have an increase prevalence of elevated fasting insulin (>/ 20 uUmL) compared to fertile, weight and age matched controls without recurrent spontaneous abortion (247). Insulin resistance is common in PCOS and may explain the increased risk of spontaneous abortion. Metformin has been found to decrease the risk of spontaneous abortion in PCOS in 2 small retrospective studies from 42-73% to 10% (248,249). However, a third study found a 35% spontaneous abortion rate in 20 conceptions in PCOS treated with metformin through 12 weeks of gestation (250). Thus there is currently no clear evidence that metformin reduces the risk of spontaneous abortion in PCOS.