This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by De Leo, V. Articles by Petraglia, F. Search for Related Content PubMed PubMed Citation Articles by De Leo, V. Articles by Petraglia, F.

Insulin-Lowering Agents in the Management of Polycystic Ovary Syndrome

Department of Pediatrics, Obstetrics, and Reproductive Medicine, Institute of Obstetrics and Gynecology, University of Siena, 53100 Siena, Italy

Correspondence: Address all correspondence and requests for reprints to: Professor Vincenzo De Leo, Institute of Obstetrics and Gynecology, University of Siena, Policlinico Le Scotte, Viale Bracci, 53100 Siena, Italy. E-mail: deleo{at}unisi.it

	Abstract

Top Abstract I. Introduction II. Definition of PCOS III. Insulin Resistance in... IV. Insulin-Lowering Strategies V. Conclusions Note Added in Proof References

 
Polycystic ovary syndrome (PCOS) is a medical condition that has brought multiple specialists together. Gynecologists, endocrinologists, cardiologists, pediatricians, and dermatologists are all concerned with PCOS patients and share research data and design clinical trials to learn more about the syndrome. Insulin resistance is a common feature of PCOS and is more marked in obese women, suggesting that PCOS and obesity have a synergistic effect on the magnitude of the insulin disorder. Hyperinsulinemia associated with insulin resistance has been causally linked to all features of the syndrome, such as hyperandrogenism, reproductive disorders, acne, hirsutism, and metabolic disturbances. Women with PCOS should be evaluated for cardiovascular risk factors, such as lipid profile and blood pressure. Modification of diet and lifestyle should be suggested to those who are obese. Several insulin-lowering agents have been tested in the management of PCOS. In particular, metformin is the only drug currently in widespread clinical use for treatment of PCOS. In a high percentage of patients, treatment with metformin is followed by regularization of menstrual cycle, reduction in hyperandrogenism and in cardiovascular risk factors, and improvement in response to therapies for induction of ovulation.

I. Introduction

II. Definition of PCOS

A. Endocrine profile

B. Reproductive abnormalities

C. Long-term health consequences

D. Summary

III. Insulin Resistance in PCOS

A. Definition and prevalence

B. Diagnosis

C. Pathogenesis of insulin resistance

D. Summary

IV. Insulin-Lowering Strategies

A. Metformin

B. Thiazolidinediones

C. Other drugs

D. Weight loss

E. Summary and clinical recommendations

V. Conclusions

	I. Introduction

Top Abstract I. Introduction II. Definition of PCOS III. Insulin Resistance in... IV. Insulin-Lowering Strategies V. Conclusions Note Added in Proof References

 
POLYCYSTIC OVARY SYNDROME (PCOS) is the most common endocrinopathy in women and the most common cause of anovulatory infertility, affecting 5–10% of the population. It is characterized clinically by some evidence of androgen excess, such as hirsutism, seborrhea, acne, elevated plasma androgen levels, or a combination of these. Its association with menstrual abnormalities and infertility leads many affected women of reproductive age to attend gynecology and infertility clinics where the syndrome is diagnosed (Table 1). In recent years it has been widely recognized that most women with PCOS have some degree of insulin resistance (1). Abnormality of insulin secretion and action has been implicated in the pathophysiology of PCOS (2). As a consequence of insulin resistance, PCOS patients often have an atherogenic lipid profile and increased incidence of cardiovascular risk factors (3). Women with PCOS have the constellation of symptoms (insulin resistance, obesity, hypertension, and dyslipidemia) defining so-called syndrome X (4). There is currently much interest in the use of insulin-sensitizing drugs in women with PCOS. In a good percentage of cases, treatment with these drugs is followed by regularization of the menstrual cycle, reduction in hyperandrogenism, reduction in cardiovascular risk factors, and improved response to therapies for induction of ovulation. This review endeavors to define insulin resistance and its possible role in the pathogenesis of PCOS. The use of insulin-lowering drugs in this disorder is summarized.

	II. Definition of PCOS

Top Abstract I. Introduction II. Definition of PCOS III. Insulin Resistance in... IV. Insulin-Lowering Strategies V. Conclusions Note Added in Proof References

The endocrine profile of women with PCOS is characterized by high plasma concentrations of ovarian and adrenal androgens, gonadotropin abnormalities, a relative increase in estrogen levels (especially estrone) derived from conversion of androgens, reduced levels of SHBG, and often high levels of prolactin (PRL) and insulin.

1. Gonadotropin secretion.
Although the pathogenesis of PCOS is still controversial, an array of plausible pathophysiologies has emerged over the last several decades of study. Inappropriate gonadotropin secretion with elevated LH and relatively low FSH secretion is typical (6).

In women with PCOS, 55–75% have a high LH to FSH ratio (6, 7, 8) due more to increased levels of LH than low levels of FSH. Administration of GnRH evokes an exaggerated LH response (9). GnRH stimulation and gonadotropin pulsatility tests indicate hyperactivity of the hypothalamo-pituitary axis. It is still unclear whether the high levels of LH depend on a higher frequency of GnRH pulses, a greater amplitude of GnRH pulses evoking an increased pituitary response to GnRH, or a combinations of these (10).

During puberty, women with PCOS showed chronobiological abnormality of LH secretion, characterized by an approximately 8-h forward shift of the LH surge from the normal nocturnal sleep period to the afternoon (11). This shift could indicate primary hypothalamic-pituitary impairment in the pathogenesis of the syndrome. However, it has been shown that in women with PCOS, recovery of the LH pulsatile pattern after inhibition by GnRH agonist closely follows the pattern observed in normal controls (12), indicating that high LH levels could be the result of increased androgen levels. Indeed, other hyperandrogenic states, such as congenital adrenal hyperplasia, may exhibit increased LH levels (13). On the other hand, the evidence that testosterone (T) administration to eugonadal female-to-male transexuals lowered LH levels (14) and that antiandrogen treatment was not always followed by a reduction in LH levels in PCOS patients argues against the hypothesis that elevated LH secretion is the consequence of hyperandrogenemia.

Similarly, the relative reduction of FSH may be explained on the basis of a higher frequency of GnRH pulses (15, 16) or a different action of circulating and paracrine factors on pituitary function. It is not yet clear whether altered hypothalamo-pituitary function is an intrinsic factor in PCOS or secondary to steroid hormone anomalies. An altered hypothalamic neuroregulation in PCOS patients has been hypothesized according to the impaired opioid and dopaminergic tonus that has been shown in PCOS. However, the administration of naltrexone (an opioid antagonist), bromocriptine (a dopaminergic agonist), or metoclopramide (a dopaminergic antagonist) induces slight changes in LH pulsatility in women with PCOS (17, 18). Whatever the cause, an altered function of the GnRH-LH axis explains some symptoms of PCOS: 1) relatively low FSH levels may lead to incomplete or inefficient follicle maturation; and 2) high levels of LH may lead to theca cell hyperplasia, favoring increased androgen secretion.

2. Hyperandrogenemia.
Hyperandrogenemia is a key feature of the syndrome; it is mainly of ovarian origin although an adrenal contribution cannot be excluded. Most, but not all, women with PCOS have high plasma levels of androgens. Androstenedione (A) and T are markers of ovarian androgen secretion, and dehydroepiandrosterone sulfate (DHEAS) is the best marker of adrenal secretion. Hyperandrogenemia is not always linked to hyperandrogenic symptoms such as acne or hirsutism; indeed, ethnic groups such as Asians show hyperandrogenemia without any skin manifestations (19, 20).

Ovarian catheterization studies (21, 22), human chorionic gonadotropin (hCG) stimulation studies (23), and suppression of gonadotropin secretion with combined estrogen and progestins (24) or GnRH analog all suggest that polycystic ovaries (PCO) overproduce androgens. The adrenal contribution, however, should not be ignored, because women with PCOS have higher plasma concentrations of T, free T, 17-hydroxyprogesterone (17OHP), A, and dehydroepiandrosterone than normal women (25). Acute administration of GnRH is followed by 3 times greater 17OHP production than in normal women (9, 25). This and other observations have led to the hypothesis that 17-hydroxylase and 17,20-lyase activities, constituents of the enzyme P450c17, are altered in PCOS. The consequence of the intrinsic dysregulation of this enzyme is a relative inhibition of 17,20-lyase with respect to 17-hydroxylase and thus an increase in the 17OHP to A ratio with respect to normal women. The increased response of 17OHP to stimulation with GnRH or hCG is one of the best known endocrine features of the syndrome (26, 27, 28). When LH is added to human thecal cell cultures, 17OHP production increases with respect to A production. Long-term cultures of replicating thecal cells from PCOS and healthy women have shown that the former produce T, progesterone, and 17OHP more abundantly than the latter, even though the culture medium was free of LH (29). The clear message from these in vitro studies is that theca cells of PCOS patients have enhanced steroidogenic potential that is not limited to 17-hydroxylase and 17,20-lyase activity. This in vitro biochemical phenotype may be the result of a stable metabolic imprint obtained in vivo or an intrinsic genetic variation.

Recent studies by McAllister and colleagues (30, 31) demonstrated that basal and forskolin-stimulated CYP17 gene transcription (the gene that encodes the cytochrome P450) is increased in PCOS theca cells. They established that increased androgen production is a stable phenotype of PCOS theca cells that not only results from preferentially increased CYP17 expression but involves the up-regulation of other steroidogenic enzymes, including CYP11A and 3ß-hydroxysteroid dehydrogenase (3ß-HSD). A comparison of 17ß-HSD activity (which converts A to T) in normal and PCOS theca cells demonstrated that androgenic 17ß-HSD activity per theca cell was not different in PCOS theca cells. It is likely that the increased production of T by PCOS theca cells is driven by increased androgen precursor production and not by altered 17ß-HSD activity (30).

In women with PCOS, altered granulosa cell function, characterized by aromatase activity, may be present. Most studies have found that aromatase activity is higher in polycystic than in normal ovaries, providing a likely explanation of the hyperestrogenism often observed in this syndrome (9, 26, 30, 31). Women with PCOS are reported to have a significantly greater response of estradiol to a single dose of GnRHa than normal women (26).

Thecal cells of PCO synthesize more androgens than those of the normal ovary, and granulosa cells have high aromatase activity that converts this large quantity of substrate into estrogens. High aromatase activity could therefore be a reason for the normal quantities of androgens found in follicular fluid from PCOS patients.

It was recently proposed that hyperinsulinemia plays a role in steroidogenesis. The relationship between steroidogenesis and insulin resistance in PCOS is discussed below.

B. Reproductive abnormalities
In clinical practice, women with PCOS present with infertility (mean incidence, 74%), menstrual irregularity (dysfunctional bleeding, 29%; amenorrhea, 51%), hyperandrogenism (69%), and virilization (21%) (32).

Anovulation is usually chronic in PCOS and is associated with infertility and dysfunctional bleeding such as oligomenorrhea or amenorrhea. Periods of regular menses are also possible. Some women who report normal menses may be anovulatory. Carmina and Lobo (33) found that approximately 21% of hyperandrogenic women with normal menses were anovulatory. The menstrual irregularity of PCOS patients typically begins at menarche and although amenorrhea may occur, the usual presentation is oligomenorrhea. The proportion of PCOS patients with regular menses is thought to increase with age, reaching about 70% at 39–41 yr (34). No longitudinal data, however, are available on this matter.

Infertility is the presenting problem for about 40% of PCOS patients (35). If pregnancy is achieved, other reproductive problems, such as miscarriage, emerge (36, 37). The relationship between PCOS and miscarriage is a complex and unresolved area. Excellent papers have been published on this topic (36, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49); consequently the subject is only outlined here. PCOS is not predictive of miscarriage, but patients who miscarry have higher plasma levels of androgens than women with ongoing pregnancies (38). The miscarriage rate in PCOS is about 30% of all pregnancies, which is double the rate for early miscarriage in normal women. The exact mechanism is unknown. High levels of LH and androgens have been regarded as a cause for poor reproductive history (50, 51).

The unfavorable endocrine environment to which ovarian follicles are exposed could be at least partly responsible for a low percentage of pregnancies, because it affects oocyte quality and luteal phase efficiency. However, oocytes of women with PCOS are nearly always normal and, when removed from their unfavorable environment, have a similar fertilization percentage to oocytes of normal women (52).

Insulin resistance in PCOS may be considered a risk factor for gestational diabetes (GD) (53). Because patients with PCOS have insulin resistance, albeit with normal glucose tolerance, they may run a higher risk of diabetes when exposed to the diabetogenic effects of pregnancy. Indeed the prevalence of GD in PCOS patients has been reported to be 40–46%. The prevalence of PCOS in women with a history of GD has variously been reported to be as high as 20% (54), 39.4% (55), 41% (56), 44% (57), and 52% (58). Interestingly, these women showed higher fasting glucose and features reminiscent of syndrome X, such as higher body mass index (BMI), higher waist-hip ratio, higher fasting insulin, higher triglycerides, and lower insulin sensitivity than women with normal previous pregnancies (58). However, some studies have failed to find a significantly higher prevalence of PCOS in women with a history of GD (59, 60).

In a longitudinal study, Paradisi et al. (61) investigated carbohydrate metabolism in pregnant women with PCOS. They used oral glucose tolerance test (OGTT) and hyperinsulinemic-euglycemic clamp to show that women who developed GD had lower insulin sensitivity than women who did not, indicating that GD is associated with impairment of insulin metabolism as early as the first trimester.

A link between insulin resistance and hypertensive disorders in pregnancy has been widely reported. Insulin resistance, obesity, and increased risk of GD in PCOS patients suggest that hypertensive complications may be common in pregnancy. Preeclampsia is reported to be more frequent in PCOS patients than in normal women (62). In a case control study, the incidence of this disorder was found to be as high as 28.5% (63). In a recent retrospective study of 99 pregnancies in PCOS patients, it was found that the relative risk for preeclampsia was 2.2. However, logistic regression analysis indicated that nulliparity was the only significant risk factor for preeclampsia and that PCOS had no predictive value (54).

C. Long-term health consequences
Insulin resistance in young, otherwise healthy women raises the question of other cardiovascular risk factors, including impaired glucose tolerance (IGT), diabetes, hyperlipidemia, hypertension, and abdominal obesity as well as increased cardiovascular disease (CVD) itself (64). Because PCOS patients tend to be obese with abdominal deposition of body fat and insulin resistance, it has been proposed that they may have other metabolic features of so-called syndrome X. This syndrome is defined by a constellation of symptoms such as insulin resistance, obesity, hypertension, and hyperlipidemia. Indeed PCOS patients tend to have higher blood pressure, triglycerides, low-density lipoprotein (LDL)-cholesterol, and total cholesterol, with lower high-density lipoprotein (HDL)-cholesterol than age-matched controls (65).

The association between PCOS and endometrial disease has been reported for many years. Few studies have addressed the possibility of an association between PCOS and breast and ovarian cancer. Multiple factors other than insulin resistance, such as obesity and hormonal imbalance, may contribute to increase the long-term risks in PCOS. This makes it difficult to evaluate the independent role of each single risk factor.

Detailed reviews on long-term health consequences of PCOS are available (66, 67, 68, 69, 70, 71, 72), and thus only a brief overview will be presented here.

1. Hypertension.
A link, independent of obesity, has been reported between hypertension and insulin resistance (73). A few studies have shown increased risk for hypertension in PCOS. Increased risk of arterial hypertension in older women with a history of PCOS has been repeatedly demonstrated (74, 75). In these retrospective cohort studies, the prevalence of treated hypertension was 3 times higher in women with a history of PCOS between the ages of 40 and 59 yr than in healthy controls (74). However, it should be acknowledged that the subjects included in that study had a history of PCOS only as documented by wedge resection and not by other clinical data. A higher prevalence of hypertension was found in a group of patients with a history of PCOS, aged 43–62 yr, matched for age and BMI with 56 controls (75). However, in a study comparing 28 patients with a history of PCOS aged 45–59 yr with 752 age- and BMI-matched controls, hypertension was diagnosed in 60% PCOS and 39% controls and the difference was not significant (76).

Normal ambulatory blood pressure was found to be similar in PCOS and normal body composition-matched women (77). However, careful comparison of blood pressure levels in young women with PCOS revealed an increase in mean and systolic blood pressure during 24-h blood pressure recording (78). A causal relation has been reported between insulin resistance and hypertension. Indeed, untreated hypertensive patients exhibited higher fasting and postprandial insulinemia than normotensive controls, and a direct correlation was found between plasma insulin levels and blood pressure (79, 80).

2. Dyslipidemia and dysfibrinolysis.
Insulin resistance is associated with an unfavorable lipid profile with low LDL and high triglyceride levels (81). Hyperinsulinemia inhibits lipolysis with a consequent increase in levels of nonesterified fatty acids. High levels of nonesterified fatty acids led to increased triglyceride levels and reduced HDL levels.

Several studies have examined the association between PCOS and dyslipidemia (65, 82, 83, 84), showing that PCOS patients have an atherogenic lipid profile with increased LDL and triglycerides and decreased HDL levels. Wild et al. (65) showed that, compared with controls, PCOS patients had serum levels of triglycerides twice as high and mean HDL levels 26% lower. Conway et al. (83) and Talbott et al. (85) showed that lean women with PCOS also had lower levels of HDL-2 subfraction than weight-matched controls. This was confirmed by Robinson et al. (86), who compared 11 lean PCOS patients with 22 BMI-matched controls and found a significant difference in HDL-2 levels, correlated with insulin resistance rather than BMI. However, not all studies found lipid abnormalities in women with PCOS, especially when they are matched to normal women for weight and body composition (87).

Insulin resistance is associated with alterations that accentuate thrombosis by increasing coagulation and inhibiting fibrinolysis (88). Plasminogen activator inhibitor-1 (PAI-1) is a potent inhibitor of fibrinolysis. Elevated PAI-1 levels have been reported in obese women (89) and lean PCOS patients (90), and a direct correlation with insulin resistance was shown. Elevated levels of fibrinogen, an independent risk factor for CVD, has been found in PCOS patients (89).

3. CVD.
Women with PCOS display a higher prevalence of cardiovascular risk factors such as obesity, hyperinsulinemia, hypertension, dyslipidemia, and dysfibrinolysis and seem to be at high risk for developing CVD.

Wild et al. (91) studied 102 consecutive pre- and postmenopausal women undergoing cardiac catheterization to investigate chest pain. Coronary artery lesions were detected in 52 patients; the other 50 had normal coronaries. The women with artery lesions had a higher incidence of hirsutism and acne than the others, and a high waist-to-hip ratio was associated with both hirsutism and coronary artery disease.

Birdsall et al. (92) performed a similar study of 143 pre- and postmenopausal women undergoing catheterization. They also performed pelvic ultrasonography to detect PCO. PCO were detected in 42% of the women. Women with PCO had more coronary artery segments with stenosis exceeding 50%, indicating a trend toward greater severity of ischemic heart disease. They found that the extent of coronary artery disease was independently associated with PCO. However PCO as detected by ultrasonography should not be confused with PCOS. Indeed it has been demonstrated that only 33% of women with PCOS showed PCO morphology on ultrasound (93).

Support for these findings also comes from a long-term follow-up study of PCOS (94). A small group of women with PCOS (n = 33) were followed up after ovarian wedge resection performed between 1956 and 1965. It was calculated that PCOS patients had a 7.4-fold greater risk of myocardial infarction than age-matched controls and increased prevalence of central obesity as well as 7-fold higher prevalence of diabetes and 3-fold higher prevalence of hypertension.

This, however, contrasts with a recent much larger retrospective study by Pierpoint et al. (95), who reported a mortality rate in 1028 women diagnosed with PCOS between 1930 and 1979. There were 59 deaths: 15 resulted from circulatory disease and six from diabetes. The standard mortality rates, both overall and from CVD, were not higher in women with PCOS than in the general female population. However, there was a significant increase in mortality from diabetes.

The same research group (96) recently investigated cardiovascular mortality in women diagnosed with PCOS before 1979 and failed to find increased mortality from CVD in these women. However, PCOS patients had a high prevalence of nonfatal cerebrovascular disease.

To evaluate vascular disease in other sites, Guzick et al. (97) measured intima-to-media thickness of the common and internal carotid arteries in 16 women more than 40 yr of age with PCOS. They found a significant increase in carotid intima-media thickness compared with that in healthy women, but no significant differences in the number of atherosclerotic plaques. The same group (98) evaluated the presence of subclinical atherosclerosis in 125 women diagnosed with PCOS between 1970 and 1990. They found that 21.6% had ultrasonographic evidence of carotid plaques compared with 15.5% of controls; 72% of PCOS patients also had plaque thickness exceeding 50% of vessel diameter compared with 0.7% of controls, suggesting that lifelong exposure to this cardiovascular risk profile may lead to premature atherosclerosis in PCOS patients.

In summary, whether PCOS is an independent risk for CVD remains unclear. The only way to establish whether PCOS is associated with high cardiovascular morbidity and mortality would be to conduct a large prospective study following women with a definitive diagnosis of PCOS for several decades.

4. Diabetes.
Insulin resistance is recognized as a major risk factor for type 2 diabetes (99). Another risk factor is pancreatic ß-cell dysfunction (100), which is also found in PCOS (101), presumably making PCOS patients at increased risk for type 2 diabetes mellitus.

Multiple factors other than insulin resistance and ß-cell dysfunction, such as obesity and family history of type 2 diabetes, may contribute to increase the diabetes risk in PCOS. This makes it difficult to evaluate the independent role of each single risk factor in the development of diabetes.

It has been reported that about 30% of obese women with PCOS have IGT. In a retrospective study Dahlgren et al. (74) observed that the prevalence of non-insulin-dependent diabetes mellitus (NIDDM) was 15% in PCOS patients compared with 2% in controls. Dunaif (102) suggested that up to 20% of PCOS patients have IGT or NIDDM by the third decade.

In a recent prospective controlled study performed in 254 women with PCOS (103), it was shown that 31% of patients had IGT and 7.5% NIDDM compared with 16% and 0%, respectively, in controls. Women aged 14–44 yr were studied from 1983–1998, and the prevalence of IGT and NIDDM was higher over 35 yr of age. By multiple regression, it was found that fasting glucose, PCOS status, waist-to-hip ratio, BMI, and age were predictors of glucose intolerance. However, in this prospective study 78% of patients had a BMI greater than 25 kg/m² and 73% were obese (BMI > 27 kg/m²). In a recent European follow-up study (104), the prevalence of diabetes was investigated in 346 PCOS patients aged 30–55 yr; 2.3% had diabetes compared with 1% of controls. When the women were divided into age groups, it was evident that more than 9% of PCOS patients aged 45–54 yr had diabetes compared with 2% of controls. The divergence with the U.S. results is probably due to the different BMI of patients. In the European study, only 44% of patients had a BMI greater than 25 kg/m² compared with 78% of patients in the U.S. study. However, both studies demonstrated that PCOS patients are at increased risk of IGT or overt diabetes during their third or fourth decade and that this risk is higher for obese than for lean patients.

Several studies have demonstrated that a family history of type 2 diabetes is frequently found in diabetic patients (105). A positive family history of diabetes is observed in a high percentage (>80%) of women with PCOS and diabetes compared with 30% among women with only PCOS (106), suggesting that a family history of diabetes may magnify the severity of insulin metabolism defects associated to PCOS. Due to the relevant pathogenetic role of insulin resistance in both PCOS and diabetes, some investigators hypothesized that among women with type 2 diabetes an higher percentage of PCOS should be found (107). Indeed the prevalence of PCOS among diabetic women seems to be 5-fold higher than normal (108). In conclusion the risk of type 2 diabetes is 5- to 10-fold higher in PCOS patients than in normal women. Obesity, insulin resistance, ß-cell dysfunction, and positive family history may contribute to the increased diabetes risk in women with PCOS. All PCOS women should be screened for glucose intolerance. Physicians need to be aware that PCOS women are at high risk for IGT and type 2 diabetes and that these abnormalities are present in both lean and obese subjects.

5. Endometrial disease.
The risk of endometrial disease is adversely influenced by several factors including obesity, unopposed estrogen, and infertility. All these factors are found in women with PCOS.

In 1970, Chamlian and Taylor (109) found that 25% of 97 cases of endometrial hyperplasia in women younger than 35 yr presenting with irregular uterine bleeding had sclerocystic ovaries (a criterion used at that time to support the diagnosis of PCOS).

Cancer incidence rates in a Mayo Clinic cohort of 1270 women with chronic anovulation, as defined by ovarian appearance consistent with PCOS and clinical evidence of chronic anovulation without hypoestrogenemia, were compared with population incidence rates (110). The relative risk for subsequent endometrial cancer associated with this syndrome was 3.1. Increased risk was also noted for premenopausal and postmenopausal cancer. In addition, 14 women reported concurrent diagnosis of chronic anovulation syndrome and endometrial cancer, consistent with a prevalence of endometrial cancer in this syndrome of approximately 1%.

A recent prospective study of 56 PCOS patients was conducted with the aim of predicting endometrial hyperplasia (111). The author found high prevalence of endometrial hyperplasia in these patients (35.7%). Of the 20 cases of endometrial hyperplasia, 12, three, and five were simple hyperplasia, complex hyperplasia, and hyperplasia with cytological atypia, respectively. Women affected were older (30–40 yr) and reported amenorrhea of 1–4 yr duration. Logistic regression analysis revealed that ultrasonographic endometrial thickness and intermenstrual interval were the only predictors of hyperplasia.

The true risk of endometrial disease in women with PCOS is difficult to ascertain. Studies have been limited to a relatively small number of cases of endometrial cancer identified specifically in PCOS. Furthermore the heterogeneous presentation of the syndrome makes it impossible to asceratin which factor (hyperinsulinemia, obesity, hormonal imbalance) has the most relevant role in the increased risk.

Clinically, it is generally accepted that in oligoamenorrheic or amenorrheic women with PCOS the induction of withdrawal bleeding to prevent hyperplasia is a prudent management.

6. Breast and ovarian cancer
Breast cancer is reported to be more common in PCOS patients (112); conversely, it has been argued that PCOS is protective against breast cancer (113).

In a large prospective study designed to examine the development of breast cancer in postmenopausal women (114), the prevalence of PCOS was found to be only 1.35%, suggesting that women presenting with the syndrome are not at increased risk for breast carcinoma. However, in a recent series of 786 women with a histological diagnosis of PCOS recorded between 1930 and 1979, breast cancer was the leading cause of death (95). Mortality was assessed from the mortality registry of deaths, and standardized mortality rates were calculated for women with PCOS and the normal population. The standardized mortality rate was 0.91 for all neoplasms and 1.48 for breast cancer [95% confidence interval (CI) 0.79–2.54] (95).

Few studies address the possibility of an association between PCOS and ovarian cancer (94, 110). Studies are limited to a small number of women with PCOS and results are conflicting.

In conclusion, the association between PCOS and breast and ovarian cancer has not been demonstrated. Although the links seem to be probable and logical, epidemiological evidence is still lacking.

D. Summary
PCOS is extremely prevalent and is considered the most frequently encountered and endocrinopathic condition. A lack of uniformity in the diagnostic criteria adds to the confusion surrounding the syndrome. The pathogenesis of PCOS is still controversial. It is likely to involve abnormalities in several systems. There has long been an association of abnormal gonadotropin secretion with this syndrome. Although the adrenal gland may contribute, hyperandrogenemia is principally ovarian in origin. During reproductive age, PCOS is associated with relevant reproductive morbidity including menstrual irregularity, anovulation, infertility, increased pregnancy loss, and complications of pregnancy. Insulin resistance is a common feature of PCOS. In the general population, insulin resistance and consequent hyperinsulinemia are associated with hypertension, dyslipidemia, dysfibrinolysis, CVD, and high risk of developing type 2 diabetes (metabolic syndrome X). It is mainly obese women with PCOS who are characterized by the presence of central obesity, insulin resistance, and dyslipidemia, which place them at a higher risk of developing diabetes as well as the possibility of CVD. However this may be true for a proportion of lean women.

Biochemical evidence regarding the potential for long-term risks of CVD is recognized. However, it remains unclear whether PCOS is associated with high CVD morbidity and mortality. Women with PCOS cluster risk factors for endometrial, breast, and ovarian cancer. However, the true incidence of endometrial, breast, and ovarian cancer in women with PCOS is not known.

	III. Insulin Resistance in PCOS

Top Abstract I. Introduction II. Definition of PCOS III. Insulin Resistance in... IV. Insulin-Lowering Strategies V. Conclusions Note Added in Proof References

 
A. Definition and prevalence
Insulin resistance has been defined as a state (of a cell, tissue, or organism) in which a greater than normal amount of insulin is required to elicit a quantitatively normal response (115). It leads to increased insulin secretion by ß-cells and compensatory hyperinsulinemia. As long as hyperinsulinemia overcomes insulin resistance, glucose levels remain normal. If ß-cell compensatory response declines, relative or absolute insulin insufficiency develops. Insulin secretion cannot keep pace with the underlying insulin resistance, which may lead to glucose intolerance and type 2 diabetes. Type 2 diabetes develops only in subjects with insulin resistance and concomitant ß-cell dysfunction. A number of abnormalities are recognized as associated with insulin resistance. Reaven (100) defined this constellation of abnormalities as syndrome X. It is intended to refer to subjects with insulin resistance, hyperinsulinemia, and dyslipidemia. Patients often have elevated triglycerides and decreased HDL-cholesterol with high blood pressure. Other frequent abnormalities are high plasma levels of PAI-1, uric acid, and fibrinogen, together with endothelial dysfunction, which make for a high risk of CVD.

In 1921 Achard and Thiers (116) first reported a relationship between hyperandrogenism and insulin metabolism in their description of "diabetes des femmes à barbe." The subsequent description in 1976 (117) of virilization in young women with severe insulin resistance led to further investigations of insulin metabolism in PCOS patients.

Both obese and lean women with PCOS have a greater insulin response to oral glucose load than healthy women (118, 119). Dunaif et al. (1) used the euglycemic glucose clamp technique to demonstrate that PCOS-associated hyperinsulinemia was caused by insulin resistance. They recruited obese and nonobese PCOS patients and showed that insulin-stimulated glucose utilization, whether expressed per kilogram total weight or per kilogram fat-free mass, was significantly lower than normal in both, indicating a type of insulin resistance that was independent of obesity and changes in body composition. With the development of easier techniques than the euglycemic clamp, such as the iv glucose tolerance test, more researchers have investigated insulin resistance in PCOS patients.

Approximately 60–70% of PCOS patients are obese, and it is well known that obesity is associated with insulin resistance. However, PCOS patients have evidence of insulin resistance beyond that of obese women in the general population. Most studies have shown that impaired insulin sensitivity is present without obesity (1, 119, 120, 121, 122); however, any degree of obesity further impairs insulin action.

Although it is universally accepted that overweight PCOS patients are insulin resistant and their insulin sensitivity is lower than that of obese non-PCOS patients, contradictory results emerged for lean women with PCOS. On the whole, European studies have failed to find insulin resistance in nonobese women with PCOS (123, 124, 125, 126, 127), whereas North American studies have either found (1, 119, 122) or not found (128) impaired insulin sensitivity in lean women with PCOS. This discrepancy may be partly due to ethnic, genetic, nutritional, and lifestyle differences.

A high percentage of women with PCOS have abnormalities of carbohydrate metabolism, such as IGT or NIDDM. Indeed, in a large prospective study by Legro et al. (103), it was shown that 38.6% of PCOS patients had either IGT (31.1%) or diabetes (7.5%) compared with 14% and 0% of controls, respectively. Lean PCOS patients had IGT in 10% of cases and diabetes in 1.5%. Similar results were reported by Ehrmann et al. (129). They found IGT in 35% and NIDDM in 10% of cases.

In conclusion, although insulin resistance is associated with obesity, it is also found in normal-weight women with PCOS. Both obese and nonobese women with PCOS seem to be more insulin resistant and hyperinsulinemic than age- and weight-matched normal women. Because insulin resistance is implicated in the pathophysiology of diabetes, women with PCOS seem to be at high risk for abnormalities of carbohydrate metabolism such as IGT and NIDDM.

B. Diagnosis
Clinical assessment of insulin resistance relies on several tests, which include determination of insulin levels, either at baseline or after OGTT, assessment of sequential plasma glucose levels after iv administration of insulin (insulin tolerance test), estimation of an index of insulin sensitivity by applying the minimal model technique to data obtained from the so-called frequently sampled iv glucose tolerance test (FSIVGTT), and measurement of in vivo insulin-mediated glucose disposal by the euglycemic hyperinsulinemic clamp procedure (for review see Ref.130). Recently, simple methods have been elaborated for assessing insulin sensitivity, such as homeostatic model assessment (HOMA) (131) and quantitative insulin-sensitivity check index (QUICKI) (132).

The most common assessment is done after oral glucose load, which represents a normal meal. The homeostatic response includes an increase in insulin secretion and insulin-dependent processes that lower glycemia. Although measurement of glycemia and insulinemia after OGTT is easy and readily available, OGTT results are difficult to reproduce and may be influenced by factors such as time of day, physical inactivity, previous carbohydrate or alcohol intake, and fasting interval before test (133). Furthermore, the results of this test must be interpreted in the context of plasma glucose levels. Indeed, any degree of hyperglycemia indicates impaired insulin secretion, further exacerbating insulin resistance and invalidating insulinemia as an index of insulin resistance. After oral glucose the increments in insulin do not depend entirely on glucose, but also on factors such as gut hormones and neural stimulation. Glycemia also changes in relation to gastric emptying and absorption.

The glucose clamp is regarded as the gold standard for assessing insulin action, and several studies have demonstrated insulin resistance in women with PCOS (Fig. 1). It is somewhat difficult to perform, requiring special equipment and trained personnel. Insulin is infused at a constant rate to achieve physiological suprabasal levels. Glycemia is monitored frequently and glucose is infused at variable rates to maintain a constant level of glycemia. When the glucose infusion rate has stabilized, this rate divided by the insulin level is defined as insulin sensitivity.

View larger version (19K): [in this window] [in a new window]   FIG. 1. Mean insulin sensitivity index (M/I, µmol/kg·min/mU·liter), glucose oxidation (black section), and nonoxidation (open section) indices expressed as µmol/kg·min/mIU·liter during the euglycemic hyperinsulinemic clamp in PCOS and controls. Indirect calorimetry was performed with a computerized flow-through canopy gas analyzer system in connection with the euglycemic clamp to reveal insulin-stimulated glucose metabolism, e.g., the rates of glucose oxidation and nonoxidation. During the euglycemic clamp, the M/I tended to be lower in lean PCOS (LPCOS) and obese PCOS (OPCOS) subjects compared with the lean controls (LC) and obese controls (OC), respectively. [Adapted with permission from L. C. Morin-Papunen et al.: Hum Reprod: 15:1266–1274, 2000 (127 ). © European Society of Human Reproduction and Embryology. Reproduced by permission of Oxford University Press/Human Reproduction.]

Most data on insulin resistance in PCOS patients have been obtained by OGTT, euglycemic clamp, or FSIVGTT. Other available techniques have not been used because of high incidence of adverse events (insulin tolerance test and insulin suppression test). Except for OGTT, techniques such as euglycemic clamp and FSIVGTT are time-, labor- and cost-intensive measures and are not feasible for large-scale screening of the population.

To avoid complex procedures or widely changing glucose levels, the HOMA focuses on basal fasting insulin and glucose levels (R_HOMA: glucose x insulin/22.5). It has been demonstrated that the correlation between HOMA and clamp-derived insulin sensitivity is surprisingly good considering the simplicity of the formula (131). A recent formula defined as QUICKI has been elaborated (132). Similar to HOMA, QUICKI is a formula based on fasting insulin and glucose values [QUICKI = 1/(log insulin + log glucose)]. It has been shown to strongly correlate with insulin sensitivity as measured by HOMA (132).

The characteristics of the most common tests for assessment of insulin sensitivity are summarized in Table 2.

View larger version (25K): [in this window] [in a new window]   FIG. 2. Nomograms indicating the relationship between waist girth and the insulin sensitivity index (M/I), obtained during a euglycemic hyperinsulinemic clamp, for different values of fasting insulin or serum triglycerides in women with PCOS. Models were constructed on 72 women with PCOS; insulin sensitivity was measured by the euglycemic hyperinsulinemic clamp. The normal ranges of insulin sensitivity were calculated from 81 nonhirsute, normally menstruating women with normal ovaries and similar BMIs and ages as the women with PCOS. Left axis, Predicted values of M/I; right axis, prediction error at that level of M/I. The nomograms also report the fifth percentile (dashed lines) of the distribution of the insulin sensitivity index for the reference group of normal women (4.9 U). Taking into account the prediction errors in a model based on fasting insulin predicted values of M/I less than 3.8 would indicate insulin resistance, whereas the M/I threshold for a model based on triglycerides would be 3.5. [Reproduced with permission from G. Gennarelli et al.: Hum Reprod 15:2098–2102, 2000 (137). © European Society of Human Reproduction and Embryology. Reproduced by permission of Oxford University Press/Human Reproduction.]

Clinical and biochemical findings usually associated with insulin resistance are reported in Table 3. These findings may help clinicians to identify women at high risk for impaired insulin sensitivity.

C. Pathogenesis of insulin resistance
As mentioned above, insulin resistance is defined as a pathological condition in which target cells fail to respond to ordinary levels of circulating insulin. At the molecular level, impaired insulin signaling results from mutations or posttranslational modifications of the insulin receptor or any of its downstream effector molecules. Insulin resistance could be accounted for by a defect in insulin binding to its receptor or to a shortage of insulin receptors; however, there is recent evidence to suggest that insulin resistance is most often due to a postbinding defect in insulin action.

1. Insulin receptor.
Insulin binds the extracellular part of its receptor, which is heterotetrameric, consisting of two - and two ß-subunits. Binding activates intracellular tyrosine kinase in the transmembrane ß-subunits. Receptor autophosphorylation triggers receptor kinase activity toward intracellular protein substrates (110), defined as insulin receptor substrates (IRSs).

The IRS family is composed of four related proteins (IRS-1 to -4) (144). These activated intermediates bind and activate other molecules, amplifying and diversifying the signal generated by insulin binding to its receptor. By activating different intermediates, insulin stimulates glucose and amino acid uptake, glycogen synthesis, lipogenesis, and mitogenesis. Any change in one of these processes (in binding of insulin to its receptor or in the postbinding signal cascade) could theoretically lead to a reduced cellular response to insulin, causing insulin resistance.

Insulin receptors are expressed in all ovarian compartments (for review see Ref.145). Insulin receptor has been demonstrated in granulosa and thecal cells and stromal tissues (146, 147). Insulin itself, IGFs, sex steroids, and other circulating factors are involved in insulin receptor expression and regulation in the ovary (145).

2. Insulin binding.
Although the pathogenesis of insulin resistance in PCOS is unclear, there is evidence that it springs from many defects, not always coexisting. A reduction in cell-surface insulin receptors has been reported in studies performed with blood cells of PCOS patients (148, 149). Although blood cells are not a classic insulin target, this finding was recently confirmed in adipocytes from lean and obese PCOS patients (150). Indeed, a marked reduction in adipocyte insulin receptor binding was found in both groups, and this appeared to be due to a reduction in insulin receptor number as opposed to a reduction in receptor affinity. However, when the study was repeated with cultured skin fibroblasts from PCOS patients, this observation was not confirmed (151). Although a significant reduction in adipocyte GLUT-4 (an insulin-regulated glucose transporter) content has been observed in PCOS, independent of obesity (152), this defect could be secondary to abnormal insulin receptor signaling.

Discrepancies in studies investigating insulin receptor binding may be due to methodological difficulties. Apart from adipocytes, insulin binding has been studied in blood cells and fibroblasts, which are not insulin target tissues. Insulin binding may depend partly on the hormonal and metabolic environment of the subjects. Hence, differences in BMI, glucose levels, insulin levels, and insulin sensitivity between studies may lead to different results. Insulin binding defects are not generally thought to play a major role in the pathogenesis of PCOS-associated insulin resistance.

3. Insulin signaling defects.
Greater attention has been paid to postbinding defects in signal transduction. In an attempt to characterize postbinding defects of insulin signaling, Dunaif (153) found that increased insulin receptor serine phosphorylation decreased its protein kinase activity. Studies of insulin receptors purified from PCOS skin fibroblasts have shown reduced insulin-stimulated receptor autophosphorylation, which appears to be the consequence of serine phosphorylation. Purification studies suggest that a factor extrinsic to the receptor (perhaps serine kinase) was responsible for serine phosphorylation (154).

Interestingly, serine phosphorylation modulates the activity of the key enzyme, P450c17, that regulates sex steroid synthesis. This suggests that a single genetic defect involving serine protein kinase could be a common cause of insulin resistance and androgen hyperproduction (155) but this hypothesis has not been confirmed. Furthermore, the defect in postbinding signaling in PCOS seems to be selective, because fibroblast cell lines from women with PCOS have significantly decreased insulin-stimulated glucose incorporation into glycogen but similar insulin-stimulated thymidine incorporation, suggesting that the defect in insulin action in PCOS is limited to the metabolic, and does not involve the mitogenic, action of insulin. In the original report, only 50% of PCOS patients exhibited a marked increase in insulin receptor ß-subunit phosphoserine (154), but PCOS patients with "normal" insulin receptors may be highly insulin resistant. This indicates that there may be other defects in the postbinding signaling that produce insulin resistance.

Indeed, a significant decrease in skeletal muscle insulin-mediated activation of IRS-1-associated phosphatidylinositol 3-kinase (PI3K) was recently demonstrated (156). Current evidence suggests that IRS-mediated activation of PI3K controls insulin-stimulated glucose transport and carbohydrate metabolism (157). Interestingly, the authors found greater expression of IRS-2 in skeletal muscle of PCOS patients than controls, suggesting a compensatory change. However insulin-mediated glucose uptake was lower in PCOS, suggesting that IRS-2-PI3K activity did not completely compensate for defective IRS-1-PI3K activity.

Lower insulin-stimulated insulin receptor tyrosine autophosphorylation was recently observed in PCOS ovaries with respect to normal ovaries (158). This could be significant, because ovaries are the main organ implicated in PCOS. Different patterns of IRS-1 and-2 expression in normal and polycystic ovaries has subsequently been demonstrated (159). IRS-1 expression is reduced in granulosa cells of ovaries from women with PCOS, and IRS-2 expression is increased in thecal cells.

Studies with adipocytes isolated from PCOS patients demonstrated reduced insulin sensitivity for induction of glucose transport, although the number of receptors and insulin affinity were unchanged (160). Treatment of adipocytes with an adenosine agonist normalized insulin sensitivity. Adenosine is thought to act as an autocrine/paracrine factor binding to G protein-linked receptors. Hence, adenosine pathways may modulate a serine protein kinase that phosphorylates insulin receptors or substrates (160, 161).

In conclusion, there may be multiple changes in the cellular pathways that mediate insulin action and signaling that underlie insulin resistance in PCOS (Fig. 3). It is therefore possible that different genetic mutations and consequently different molecular abnormalities result in the same phenotype.

View larger version (24K): [in this window] [in a new window]   FIG. 3. Multiple defects in the cellular pathways that mediate insulin action and signaling may underlie insulin resistance in PCOS. Increase in serine phosphorylation and decrease in tyrosine phosphorylation and in IRS expression and activation may be involved in the pathogenesis of PCOS-associated insulin resistance.

Hyperinsulinemia may increase androgen production in PCOS by stimulating the ovaries directly, or indirectly through stimulation of LH secretion and inhibition of IGF binding protein (IGFBP) and SHBG synthesis and secretion. Finally, insulin may also stimulate adrenal androgen secretion (Fig. 4).

View larger version (25K): [in this window] [in a new window]   FIG. 4. Pathogenesis of insulin resistance and role of hyperinsulinemia in the pathophysiology of PCOS. Genetic predisposition, obesity, and body fat location may have independent effects on insulin sensitivity. Hyperinsulinemic insulin resistance could have a central role in the pathogenesis of PCOS. Hyperinsulinemia may increase androgen levels by stimulating ovarian steroidogenesis and inhibiting IGFBP and SHBG synthesis and secretion. A role for insulin on adrenal steroidogenesis has been hypothesized.

Studies on thecal cells are more interesting and show that insulin acts as a cogonadotropin in steroidogenesis (165). Insulin has been shown to stimulate proliferation of thecal cells (166), to increase LH-stimulated androgen secretion (167, 168, 169), to increase P450c17 mRNA levels (170), to up-regulate LH receptors (171), and to up-regulate ovarian IGF-I receptors (145).

Although in vitro studies have generally shown major effects of insulin on ovarian steroidogenesis, in vivo studies tend to be divergent. The results of studies involving infusion of insulin in vivo followed by measurement of plasma levels of androgens have been both negative (172, 173, 174, 175) and positive (176, 177, 178, 179). The reason is probably related to the brief duration of the studies; a few hours is not long enough to detect insulin-induced steroidogenesis. Studies in which circulating levels of insulin were reduced by administration of diazoxide (180), somatostatin (181), acarbose (182), metformin (183, 184), and troglitazone (185) are more convincing. As illustrated in Section IV, reduction of hyperinsulinemia may be followed by a decrease in basal plasma levels of androgens in response to administration of GnRH (183, 186, 187) or hCG (28).

b. Insulin and gonadotropin.
There is evidence to suggest that insulin of pancreatic origin penetrates the brain (188), and insulin receptors have been identified in various brain regions, particularly the hypothalamus (189). In mice with neuron-specific disruption of the insulin receptor gene, a reduction in circulating LH and enhancement of LH response to GnRH have been observed (190). This suggests that insulin signaling in the brain could be essential for normal regulation of the hypothalamic-pituitary-ovarian axis. Insulin receptors have also been identified in the pituitary (189), and insulin has been demonstrated to modulate pituitary activity in vitro (191). In the well-known study of Adashi et al. (192), insulin was shown to stimulate both basal and GnRH-stimulated release of LH and FSH in rat pituitary cells.

In vivo studies, however, have not completely clarified the role of insulin in gonadotropin secretion. Those based on insulin infusion have failed to demonstrate changes in gonadotropin responses to GnRH or pulsatile gonadotropin release (178). Correlation studies suggest that LH levels and pulse amplitudes are inversely related to insulin levels and to the degree of insulin resistance (122, 193). Insulin could therefore inhibit, rather than stimulate, gonadotropin secretion. Indeed, obese hyperinsulinemic women with PCOS have decreased LH levels compared with lean women (193, 194).

Studies in which insulin levels were reduced by means of drugs are more univocal. As illustrated below, insulin-lowering drugs, according to some but not all authors, bring about a reduction in basal levels of LH (195, 196, 197, 198) and of the LH response to GnRH (183, 186, 187). However, it should not be forgotten that spontaneous or induced ovulations rapidly lower LH levels and may be a confounding factor.

c. Insulin and IGF- and sex hormone-binding proteins.
Insulin regulates androgen metabolism, not only affecting synthesis and secretion, but also indirectly regulating circulating levels of SHBG, which has high affinity for sex hormones. As plasma concentrations of SHBG become lower, the free or bioavailable androgen fraction becomes greater.

In vitro studies indicate that insulin suppresses SHBG production by cultured hepatoma cells (199), and in vivo studies indicate an inverse correlation between insulinemia and SHBG plasma levels in hyperandrogenic women and also in the general population (200, 201, 202, 203). A reduction in hyperinsulinemia in PCOS patients leads to a significant increase in circulating levels of SHBG, which is certainly a main reason for improvement of hyperandrogenism in response to therapy (185, 202, 204).

Insulin also inhibits liver production of IGFBP-1, which in turn leads to an increase in the free fraction of IGF-I. Insulin suppresses IGFBP-1 gene transcription. An inverse relationship between fasting insulin and IGFBP-1 has been demonstrated in PCOS patients and the general population (205, 206, 207). Insulin reduces not only hepatic synthesis of IGFBP-1 but also the intraovarian pool (166).

Several studies have found a significant increase in the IGF-I to IGFBP-1 ratio in women with PCOS (208, 209, 210). As a consequence, increased bioavailability of IGF-I to thecal tissue may subserve a cogonadotropin role, inducing hyperandrogenism by autocrine and paracrine mechanisms; indeed, IGF-I has been shown to stimulate estrogen production by granulosa cells (211) and to act synergistically with FSH and LH in controlling granulosa cell aromatase levels (212, 213). IGF-I synergizes with LH to stimulate androgen production (165, 167), probably via its receptors on thecal cells (165).

Furthermore IGF-I, like insulin, could indirectly control ovarian steroidogenesis by affecting hypothalamic-pituitary function. Indeed, IGF-I has been shown to positively regulate GnRH gene expression (214, 215) and to increase basal and GnRH-stimulated pituitary gonadotropin release (216). Insulin-lowering treatment increases IGFBP-1 plasma levels, thereby leading to a reduction in the IGF-I to IGFBP-1 ratio and reducing the free fraction of IGF-I available in peripheral tissues (217, 218).

d. Insulin and the adrenal.
As mentioned above, adrenal hyperandrogenism has been considered a characteristic feature of PCOS. Because insulin resistance is a main component of the endocrine pattern of PCOS, many studies have evaluated the hypothesis that adrenal hyperandrogenism could be secondary to hyperinsulinemia. The key enzyme of androgen synthesis, P450c17, is expressed both in the gonads and adrenals and is encoded by the same gene (219); therefore, the same factor (i.e., insulin) probably regulates enzyme activity at both sites.

Physiological concentrations of insulin and IGF-I were found to increase P450c17 mRNA levels irrespective of ACTH, in cultured adrenocortical cells (170, 220, 221). Insulin and IGF-I also increased type II 3ß-HSD mRNA levels and had less effect on 21-hydroxylase mRNA levels (170).

Alternatively, some in vitro studies have shown a stimulatory effect of insulin on adrenal steroid production (222), and others have shown inhibitory effects (223). Nor do the findings of in vivo studies agree. Indeed, an inverse correlation has been shown between plasma levels of insulin and DHEAS in human males (224), and certain epidemiological studies have shown low levels of DHEAS in states characterized by insulin resistance and hyperinsulinemia, such as obesity (225). Moreover, acute insulin infusion in healthy women is reported to cause a drop in DHEAS levels (175). Short-term insulin infusion resulted in an increased response of some steroid intermediates to ACTH stimulation and suggested that insulin may drive an increase in 17-hydroxylase and relative impairment of 17,20-lyase activity (226).

Hyperinsulinemia in women with PCOS affects adrenal androgen production; the response of A and 17OHP to ACTH was greater in hyperinsulinemic women than in normoinsulinemic women with PCOS (227). Studies in which circulating levels of insulin are reduced by administration of metformin (187, 228) confirmed that insulin has a role in regulating adrenal steroidogenesis. Metformin administration was followed by a significant reduction in the response of 17OHP, T, and A to ACTH (228), indicating reduced adrenal steroidogenesis in response to lower circulating insulin levels. Pharmacological reduction of insulin levels was associated with a significant decrease in 3ß-HSD activity in C21 steroids (which converts 17OHP into 17OH pregnenolone), in 17ß-HSD activity (which converts T to A), and an increase in 17,20-lyase activity in the D4 pathway (which converts A to 17OHP) (187). No changes in 3ß-HSD activity in C19 steroids and 17,20-lyase activity in the D5 pathway were observed. These observations may indicate a stimulating effect of insulin on 3ß-HSD and 17ß-HSD activity and an inhibiting effect on 17,20-lyase activity.

As in the case of the ovaries, the major abnormalities in adrenal steroid secretion seem to be dysregulation of 17-hydroxylase and 17,20-lyase activities. As in the ovaries, insulin may largely affect the regulation of adrenal steroidogenesis.

5. Role of obesity.
Acquired factors seem to play an important role in the pathogenesis of PCOS-associated insulin resistance. Indeed, obese PCOS patients are more insulin resistant than lean ones. Although the association of obesity and insulin resistance is universally accepted, the mechanisms by which increased adipose tissue causes insulin resistance remain unknown. It is significant that the sites of adiposity are not equal in this regard. Intraabdominal fat deposits (central obesity) are much more strongly associated with impaired insulin sensitivity. Central obesity is thought to induce insulin resistance by expressing and secreting several peptide hormones and cytokines. Increased production of free fatty acids (FFA) may inhibit insulin clearance and induce defects in cell uptake of glucose and glycogen synthesis (229, 230).

a. Emerging role of TNF-.
Adipocytes are well known for their relevant role as energy storage depots. New data have established an additional role for adipocytes, that of secretory cell. Adipocytes secrete peptide hormones and cytokines, including TNF-, which helps maintain hemostasis.

Expression of the cytokine TNF- is greater in adipose tissue and muscle of obese animals. The degree of TNF- expression is correlated positively with the degree of obesity (231). Circulating levels of this factor are elevated in obese subjects and decrease with weight reduction (232).

TNF- has many effects on adipocyte function including inhibition of lipogenesis and stimulation of lipolysis. TNF- exerts a direct effect on insulin signaling by stimulating serine phosphorylation of IRSs with consequent impairment of cellular response to insulin (233). Interestingly, high circulating TNF- has been reported in normal-weight PCOS patients and even higher levels in obese PCOS patients (234, 235). Thus, increased circulating TNF- may be involved in the pathogenesis of insulin resistance associated with PCOS.

D. Summary
Hyperinsulinemic insulin resistance is characteristic of many, if not all, women with PCOS. Women with this syndrome also display a high prevalence of glucose intolerance and type II diabetes. Generally, the euglycemic hyperinsulinemic clamp is considered the gold standard for the measurement of insulin sensitivity in vivo. Because of its cost, time, and technical demands, other simpler methods, such as HOMA and QUICKI, may be used in clinical practice. Some clinical and biochemical findings may be useful to detect the presence of insulin resistance.

Pathogenesis of insulin resistance in PCOS is still a matter of debate. Defects in insulin binding to its receptor or, most probably, defects in downstream effectors of the insulin receptor may be the molecular sites of insulin resistance in PCOS. Obesity and body fat location have important independent effects on insulin sensitivity. Adipocyte-derived factors (FFA and TNF-) seem to directly affect the cellular action of insulin.

Hyperinsulinemic insulin resistance is considered to have a central role in the pathogenesis of PCOS. Hyperinsulinemia may lead to inappropriate gonadotropin secretion and to adrenal hyperandrogenism. However, experimental evidence suggests that hyperinsulinemia could produce hyperandrogenism by directly increasing ovarian androgen synthesis and reducing SHBG plasma levels.

	IV. Insulin-Lowering Strategies

Top Abstract I. Introduction II. Definition of PCOS III. Insulin Resistance in... IV. Insulin-Lowering Strategies V. Conclusions Note Added in Proof References

 
On the basis of the theory that insulin resistance and hyperinsulinemia may be a relevant contributor to the pathophysiology of PCOS, it has been hypothesized that insulin-lowering agents, by reducing hyperinsulinemia, might improve endocrine and reproductive abnormalities with PCOS.

A. Metformin
There is now a large body of data documenting the clinical efficacy of metformin in the treatment of PCOS-associated insulin resistance. Metformin is an "old" drug (Fig. 5), mainly used to lower blood sugar in NIDDM. Its mechanism of action is still not entirely understood. Metformin and phenformin were introduced in 1957. Phenformin was withdrawn from clinical use in many countries in the late 1970s, when an association with lactic acidosis was recognized. Metformin became available for use in the United States in 1995. It is administered orally and improves insulin sensitivity that is impaired in NIDDM. Its efficacy is considered similar to that of sulfanilylurea.

View larger version (17K): [in this window] [in a new window]   FIG. 5. Chemical structures of metformin and troglitazone.

b. Effect on peripheral glucose utilization.
Metformin increases insulin-stimulated glucose utilization (estimated by means of the hyperinsulinemic clamp) by up to 50% in subjects with NIDDM (246) or normoglycemic insulin resistance (247). This effect has been predominantly attributed to an increase in nonoxidative glucose metabolism (240) with glucose oxidation being less affected. Nonoxidative glucose metabolism includes storage as glycogen, conversion to lactate, and incorporation into triglycerides (Table 4).

c. Effect on insulin levels and insulin receptor.
A significant reduction in insulin and proinsulin levels in lean and overweight patients with NIDDM has been reported (251, 252). This effect may be secondary to the glucose-lowering effect of metformin. A report of the Biguanide and the Prevention Risk of Obesity has indicated that metformin significantly reduces fasting insulin levels in nondiabetic subjects. After 12 months, 164 patients taking metformin showed a reduction in plasma insulin of about 36 pmol/liter (253). Insulin binding to its receptor is reduced in NIDDM. Attempts have been made to determine the effect of metformin on the extent of insulin binding. Some studies have reported improved binding of insulin on erythrocytes and monocytes in healthy subjects and in obese and lean patients with NIDDM (254).

Other studies, however, found no change in either the number of insulin receptors on erythrocytes and monocytes or the affinity of insulin for its receptor (246, 255). Moreover, the extent of insulin receptor binding did not appear to be correlated with the clinical and metabolic response to metformin, indicating that these effects are probably mediated at the intracellular level (256). Indeed, it has recently been shown that metformin enters the cell and directly stimulates the tyrosine kinase activity of the intracellular portion of the ß-subunit of the insulin receptor (257).

d. Side effects and toxicity.
Reversible gastrointestinal side effects occur in about 30% of patients taking metformin. These effects include diarrhea, nausea, abdominal discomfort, anorexia, and a metallic taste in the mouth. These effects are not severe and can be avoided by taking metformin with food or by commencing therapy with a low dose. The newly available sustained-release formulations seem to be associated with a reduction in gastrointestinal side effects (258).

Reduced gastrointestinal folic acid and vitamin B12 has been reported, although without clinical symptoms (259). The worst toxic effect of metformin is probably lactic acidosis, which is fortunately not common. The US Food and Drug Administration recently reported a rate of five cases/100,000 treated patients (260). Lactic acidosis due to metformin therapy is associated with a mortality rate of about 50%. Mortality increases with the degree of renal impairment.