Hyperandrogenism and Hyperinsulinemia

Hyperinsulinemia and hyperandrogenism represent a tangled web that reaches its greatest complexity in women with polycystic ovary syndrome (PCOS). Researchers are beginning to separate the strands of the relationship between the two, but the pattern is still not clear. Initial data seem to suggest that the hyperinsulinemia, rather than the hyperandrogenemia, holds perhaps the gravest implications for long-term complications of the syndrome, such as cardiovascular disease and development of non - insulin-dependent diabetes mellitus (NIDDM). But the data supporting these long-term complications in women with PCOS are still scant. As gynecologists, we have traditionally focused on the hyperandrogenic aspects of this syndrome: hirsutism, dysfunctional uterine bleeding, and infertility. But as we identify the long-term sequelae, our primary concern for these patients may become improving insulin sensitivity. Because they frequently develop symptoms at the time of menarche, they represent an ideal target group for early intervention.

This chapter will focus mainly on insulin resistance in PCOS, but other causes of hyperandrogenism and hyperinsulinemia also will be discussed.

Insulin is the primary glucoregulatory hormone and is secreted by the beta cells of the pancreas. Its primary target tissues are skeletal muscle, liver, and adipose tissue. Of these, skeletal muscle is the largest target for insulin action. In the liver, insulin promotes glycogen formation and inhibits gluconeogenesis and glycogenolysis. In fat and muscle, insulin promotes glucose utilization and storage. Insulin is a potent growth factor, and its effects on peripheral tissues are pleiotropic and potentially not all favorable. In high doses, it may be directly atherogenic or may alter circulating lipids and the coagulation cascade in a negative manner.

Insulin resistance is defined as a subnormal target tissue response to a given amount of insulin.¹ This can result in a compensatory increase in insulin secretion by the pancreas, which initially can maintain euglycemia. Insulin resistance may, however, be a progressive disorder with either greater resistance to the effects of insulin at the target tissue or a decrease or inability of beta-cell function to continue to compensate with increased production of insulin. This can lead to elevated circulating levels of glucose, impaired glucose tolerance, and eventually frank diabetes mellitus. Fasting hyperinsulinemia may be a predictor of insulin resistance, at least in the earlier stages of insulin resistance. But exhaustion of the beta cells of the pancreas may lead to declining circulating insulin levels. Diabetics may have marked insulin resistance in the peripheral tissues, combined with subnormal or normal circulating insulin levels. Therefore, although in most cases hyperinsulinemia indicates the presence of insulin resistance, insulin resistance may not always be accompanied by elevated circulating levels of insulin. However, peripheral insulin levels are quite variable, and insulin sensitivity can be much more precisely quantified by dynamic tests of insulin action, rather than a spot check of insulin levels.

Insulin resistance has a wide variety of etiologies, both congenital and acquired, and is found in a variety of normal physiologic states, such as pregnancy. A summary of potential etiologies is found in Table 1. Though many authors had previously noted the interconnection of obesity, diabetes, and cardiovascular disease, the insulin resistance syndrome as it is now commonly referred to was initially expounded by Reaven in 1988. This syndrome (also known as syndrome X) was characterized by glucose intolerance (with hyperinsulinemia), lipid abnormalities (increased very-low-density lipoprotein and triglycerides, and decreased high-density lipoprotein [HDL]), and hypertension.² These changes were believed to increase markedly the risk for coronary artery disease in affected subjects. These effects, especially in the presence of insulin resistance, may be even more pronounced in women. Diabetic women have nearly a twofold higher mortality rate from coronary artery disease than diabetic men.³ Subsequent research has expanded the number of metabolic abnormalities associated with the insulin resistance syndrome.⁴ A list of these can be found in Table 2.

TABLE 1. Causes of Insulin Resistance

  Defects intrinsic to target cells

  Mutations of the insulin-receptor gene
  Defects in other genes important for insulin action (putative)

  Glucose transporters
  Substrates for insulin-receptor kinase or signaling intermediates
  Cellular inhibitors of insulin-receptor kinase

  Secondary factors affecting target cells

  Abnormal physiologic states

  Stress (e.g, fever, sepsis)
  Fasting or starvation
  Uremia
  Cirrhosis
  Ketoacidosis
  Obesity
  Diabetes or hyperglycemia

  Normal physiologic states

  Puberty
  Advanced age
  Pregnancy

  Specific hormonal or metabolic factors

  Glucocorticoids (e.g., Cushing's syndrome)
  Growth hormone (acromegaly)
  Catecholamines (e.g., pheochromocytoma)
  Glucagon (e.g., glucagonoma)
  Thyroid hormone (thyrotoxicosis)
  Hyperinsulinemia (e.g., insulinoma)
  Hyperglycemia (diabetes)
  Free fatty acids (nonesterified fatty acids)
  Adenosine
  Perhaps islet amyloid polypeptide (amylin)

  Autoantibodies to the insulin receptor

(Moller DE, Flier JS: Insulin resistance: Mechanisms, syndromes, and implications. N Engl J Med 325:938,1991)

TABLE 2. Insulin Resistance Syndrome

  Insulin resistance
  Compensatory hyperinsulinemia

  Normal glucose tolerance
  Impaired glucose tolerance
  Mild type II diabetes mellitus*

  Central obesity
  Hypertension
  Dyslipidemias

  Increased fasting triglyceride concentration
  Decreased high-density lipoprotein cholesterol concentration
  Small, dense, low-density lipoprotein cholesterol particlest
  Large, postprandial, triglyceride-rich lipoprotein particles^†

  Coronary artery disease
  Hyperuricemia^†
  Increased plasminogen activator inhibitor type l^†
  Decreased plasminogen activator activity^†
  Microvascular angina (in absence of large vessel coronary artery disease)^†

*Insulin concentrations are no longer elevated in more severe type II diabetes mellitus (e.g., fasting plasma glucose levels approximately >180 mg/dL).
^† These factors are recent additions³ and will require further studies for confirmation.
(Davidson MB: Clinical implications of insulin resistance syndromes. Am J Med 99:420,1995)

Not all women who are insulin resistant or have the insulin resistance syndrome are hyperandrogenic. These women may be a subset or a unique entity of syndrome X. The association between insulin resistance and hyperandrogenism was first noted in 1921 by Archard and Thiers,⁵ who referred to the “diabetes of bearded women.” Several syndromes of hyperandrogenism and insulin resistance have since been well characterized. The rare causes tend to involve marked hyperinsulinemia and a congenital onset.

Rare Causes of Hyperandrogenism and Insulin Resistance

TYPE A AND TYPE B SYNDROME.

Kahn and colleagues⁶ subsequently described a group of patients with insulin resistance, acanthosis nigricans, and hyperandrogenism. They elucidated two types of hyperandrogenic insulin-resistant syndromes: type A syndrome, due to genetic mutations in the insulin receptor; and type B syndrome, due to autoantibodies to the insulin receptors.⁷ The presentation in both syndromes is similar, although women with type B syndrome may be affected by other autoimmune disorders.

LEPRECHAUNISM.

Leprechaunism has been characterized by severe intrauterine growth retardation; dysmorphic facies; a large, protuberant abdomen; large hands, feet, and genitalia; hypertrichosis; acanthosis nigricans; and decreased subcutaneous fat.⁸ Reddy and Kahn⁹ also noted a functional abnormality in the epidermal growth factor receptor in addition to the insulin receptor and have proposed that there may be a more fundamental defect beyond the insulin receptor in these patients.

RABSON-MENDENHALL SYNDROME.

Rabson-Mendenhall syndrome is similar in phenotype to leprechaunism, but patients display early dentition, prognathism, and thick fingernails. Pineal hyperplasia has also been frequently noted.¹⁰ Insulin receptor abnormalities may be responsible for this syndrome.¹¹

LIPOATROPHY.

Lipoatrophy, also known as Kobberling-Dunnigan syndrome or reverse partial lipodystrophy, has primarily been reported in women. There may be two phenotypes involving loss of subcutaneous fat: one with with lean, muscular limbs, with fat loss sparing the face and trunk; and another with truncal involvement. In addition to the hyperandrogenic insulin resistance, hyperlipidemia and hyperuricemia have been reported.¹² Unlike the other congenital causes already discussed, lipoatrophy may present in adolescent girls at the time of puberty.

In contrast to the above rare entities, PCOS is encountered in some fashion by most practicing physicians every day. Several of the previously mentioned syndromes have been attributed to abnormalities in the insulin receptor with a wide range of possible phenotypes. These syndromes are extremely rare in women, and they tend to have a earlier and severer presentation than PCOS (except for lipoatrophy). PCOS tends to have a perimenarchal onset, and insulin resistance is rarely associated with wasting syndromes and diabetic ketoacidosis, as are many of the above-mentioned entities. However, PCOS displays a wide range of phenotypes and probably represents a heterogeneous group of disorders.

Polycystic Ovary Syndrome: Definition

PCOS remains a diagnosis of exclusion. The cause is unknown, and the exact definition of the syndrome is frequently author-dependent. The current recommended diagnostic criteria for PCOS based on a 1990 NIH-NICHD conference on PCOS¹³ (reconfirmed in a 1995 Serono Symposia on PCOS¹⁴) are chronic anovulation and hyperandrogenism with exclusion of other etiologies (i.e., nonclassic congenital adrenal hyperplasia, hyperprolactinemia, Cushing's syndrome, and androgen-secreting tumors). It is important to exclude these other etiologies, especially hypercortisolism, as this may be a secondary cause of insulin resistance. Although there have been no specific population-based studies, a 5% to 10% prevalence of this disorder in women of reproductive age may be a reasonable conservative estimate. This is based on an upper limit in studies of the prevalence of polycystic ovaries, which found that approximately 20% of self-selected normal women had polycystic ovary morphology on ovarian ultrasound, although a substantial proportion had no identifiable endocrine abnormality.¹⁵

Polycystic Ovary Syndrome and Insulin Resistance

Givens and associates,^16,17,18 from the University of Tennessee in Memphis, have reported on multiple PCOS kindreds, showing affected members in several generations. These studies were the first to reveal some of the severe metabolic sequelae that may accompany the syndrome, including diabetes mellitus, insulin resistance, lipid abnormalities, hypertension, and arteriosclerosis (part of a larger sample pedigree is shown in Fig. 1). This group subsequently reported on a correlation between hyperandrogenism and hyperinsulinemia in women with PCOS, with significant basal and stimulated hyperinsulinemia compared to controls.¹⁹ Subsequent studies by others have demonstrated insulin resistance in almost all women with PCOS, both lean and obese.^20,21,22 The study by Dunaif and colleagues²² was the first systematic assessment of glucose tolerance in women with PCOS. Twenty percent of the obese women with PCOS had impaired glucose tolerance or frank NIDDM according to National Diabetes Data Group Criteria. Insulin resistance in women with PCOS approaches that found in NIDDM. The mechanism of insulin resistance in PCOS may involve abnormalities in signal transduction, although it may not involve genetic abnormalities of the insulin receptor, as in the rare syndromes mentioned earlier. No insulin receptor mutations have been detected by direct sequencing of genomic DNA in women with PCOS²³; however, they have been found to have defects in insulin receptor phosphorylation.^24,25

Despite the proposed abnormality in signal transduction, the exact relationship between hyperandrogenism and hyperinsulinemia in women with PCOS is unknown. Theories have proposed that hyperinsulinemia causes hyperandrogenism, that hyperandrogenism causes hyperinsulinemia, or that some unknown third defect is responsible for both phenomena.

Hyperandrogenism Causes Hyperinsulinemia

One would expect all men to be insulin resistant if hyperandrogenism caused insulin resistance, as the average circulating testosterone level in men is approximately 10 times that of normal women, and even several-fold higher than in women with PCOS. Yet that is not the case, and there is evidence to suggest that the relationship between circulating androgens and insulin resistance may be sexually dimorphic. Potent androgens may, however, produce mild insulin resistance in women. Women receiving oral contraceptives containing “androgenic” progestins can experience a worsening in glucose tolerance, as can subjects receiving anabolic steroids.^26,27 Prolonged testosterone administration to female-to-male transsexuals that produced circulating testosterone levels in the normal male range resulted in insulin resistance in euglycemic clamp studies.²⁸ The role of adrenal androgens, such as dehydroepiandrosterone (DHEA) and DHEA-S, in promoting insulin resistance is more complex. These androgens are often elevated in women with PCOS. Buffington and colleagues²⁹ proposed that androgens have differential effects on insulin action, with testosterone worsening insulin sensitivity and DHEA improving it. Studies in which DHEA or DHEA-S has been administered to humans have had divergent results. Administration of supraphysiologic amounts of DHEA (which also result in testosterone elevations, because DHEA is a testosterone precursor) has produced mild hyperinsulinemia in postmenopausal women.³⁰

If hyperandrogenism caused insulin resistance, amelioration of the hyperandrogenism would be expected to improve insulin sensitivity. This has not been clearly demonstrated in women with PCOS. Prolonged androgen suppression achieved with an agonist analog of gonadotropin-releasing hormone either has not improved insulin sensitivity,³¹ or has resulted in only mild improvements in insulin sensitivity in PCOS during androgen suppression.³² Antiandrogen therapy has also failed to produce significant improvements in insulin resistance.³³ In summary, the modest hyperandrogenism characteristic of PCOS may contribute slightly to the insulin resistance. Additional factors are necessary to explain the degree of insulin resistance found in women with PCOS.

Hyperinsulinemia Causes Hyperandrogenism

More support in the literature may be found for the premise that hyperinsulinemia causes ovarian hyperandrogenism. Poretsky³⁴ has pointed out, however, the apparent contradiction that other primary target tissues remain resistant to insulin while the ovary appears sensitive. Insulin can be shown experimentally to have a variety of direct actions on steroidogenesis in humans. Insulin can stimulate ovarian estrogen, androgen, and progesterone secretion in vitro. Recent data suggest that physiologic insulin levels enhance androgen production from the granulosa cells of polycystic ovaries and may act synergistically with luteinizing hormone.³⁵

Despite the diverse effects of insulin on ovarian function in vitro, it has been more difficult to demonstrate similar actions in humans in vivo. There are several other well-documented in vivo conditions of hyperinsulinemia that are not consistently associated with hyperandrogenism. These include simple obesity and NIDDM. Simple obesity has not been consistently associated with abnormal circulating androgens. Elevated testosterone levels were not noted in women with NIDDM compared to controls, although both groups were postmenopausal.³⁶ A better model are the Pima Indians, a population with a high prevalence of hyperinsulinemia. Reproductive-age Pima Indian women who are hyperinsulinemic are not hyperandrogenic; in fact, no difference in androgen levels was noted between normoinsulinemic and hyperinsulinemic women.³⁷ Relatively physiologic levels of insulin only slightly increase circulating androgen levels in normal women.³⁸ Acute insulin infusions decrease DHEA-S levels in women, suggesting that insulin is a negative modulator of adrenal androgen metabolism.³⁹ Insulin can directly decrease hepatic sex-hormone - binding globulin (SHBG) production, explaining the frequently observed inverse correlation between peripheral insulin and SHBG levels.⁴⁰ Insulin, rather than sex steroids, appears to be the major regulator of SHBG production.⁴¹ Decreased circulating SHBG results in increased free and thus bioavailable androgens. This would augment ovarian hyperandrogenism in women with PCOS.

The Unknown Third Factor

A third potential etiologic category, an unknown factor contributing to both hyperandrogenism and hyperinsulinemia, remains by definition largely unexplored. The complexity and heterogeneity of the relationship between hyperandrogenism and hyperinsulinemia suggests some basic, but as yet unidentified, factor. It is anticipated that molecular genetic investigation of PCOS women and their families, perhaps combined with systematic linkage analysis, may unearth genetic factors that determine the phenotype of PCOS.^42,43

Polycystic Ovaries

Polycystic ovaries are characterized by multiple subcapsular follicular cysts (2 to 10 mm) and a thickened capsule, and are enlarged both due to the preceding factors and to increased central stroma. They have a characteristic appearance on transvaginal ultrasound according to the frequently cited criteria of Adams and co-workers,⁴⁴ which includes the presence of eight or more peripheral follicular cysts (less than or equal to 2 to 10 mm) with increased central ovarian stroma (Fig. 2). Polycystic ovaries represent a final common phenotype of a wide variety of etiologies or, as Givens⁴⁵ stated so succinctly, they are “a sign, not a diagnosis.” Although polycystic ovaries have been associated with a wide variety of apparently unrelated conditions (Table 3), many practitioners use the presence of polycystic ovaries as a screening test for PCOS. However, their presence also can be found in normoinsulinemic, normoandrogenic, cycling women. Hyperandrogenism has frequently been associated with polycystic ovaries, although there are many other pathologic entities without overt hyperandrogenism that may also be associated with polycystic ovaries. Recently, women with glycogen storage disease (types IA and III), a syndrome of insulin resistance without hyperandrogenism, were noted to have polycystic ovaries.⁴⁶ All females, including children, were noted to have polycystic ovaries. Both prepubertal and postpubertal females were noted to have normal levels of circulating androgens. This association suggests that in certain instances insulin resistance alone can cause polycystic ovaries.

TABLE 3. Syndromes or Disease Entities That Have Been Associated with Polycystic Ovaries

  Hyperandrogenism
  Steroidogenic enzyme deficiencies (CAH, aromatase deficiency, etc.)
  Androgen-secreting tumors

  Ovarian
  Adrenal

  Exogenous androgens

  Anabolic steroids
  Transsexual hormone replacement

  Hyperandrogenism and Insulin Resistance
  Congenital

  Type A syndrome
  Type B syndrome
  Leprechaunism
  Lipoatrophic diabetes
  Rabson-Mendenhall
  Polycystic ovary syndrome (?)

  Acquired

  Cushing's syndrome

  Insulin Resistance
  Glycogen storage diseases
  Other
  Central nervous system

  Trauma/lesions
  Hyperprolactinemia

  Nonhormonal medications

  Valproate

  Hereditary angioedema
  Bulimia
  Idiopathic

  (Includes normoandrogenic women with cyclic menses)

CAH = congenital adrenal hyperplasia

Insulin resistance is characteristic of PCOS, or otherwise stated, the syndrome of hyperandrogenism and menstrual irregularity; however, it is not found in endocrinologically normal women with polycystic ovary morphology, or in hyperandrogenic women with ovulatory cycles.^47,48,49 Additionally, women who have all the endocrinologic manifestations of the syndrome, including insulin resistance, can have normal ovarian morphology on ultrasound.⁵⁰ Polycystic ovaries are a puzzling phenomenon: their presence or absence may not be relied upon to make the diagnosis of hyperinsulinemic hyperandrogenism.

Body Fat Distribution

When discussing women with PCOS, obesity is always a potential confounding variable, and it is difficult to separate out the independent and interactive effects of hyperandrogenism, obesity, and insulin resistance in these women. Simple obesity can result in acquired insulin resistance after a weight gain of as little as 15% above ideal body weight.⁵¹ However, the pattern of fat distribution rather than body mass index seems to be most predictive of morbidity and mortality. Centrally distributed fat, giving some women an apple shape, is referred to as an android fat distribution. This has been associated in women with an increased risk of cardiovascular disease, especially myocardial infarction,⁵² and with mortality from all causes.⁵³ Fat distributed mainly in the hips, giving some women a pear shape, is referred to as gynecoid fat distribution. This is best detected by obtaining a circumference both at the umbilicus and at the hips around the level of the symphysis pubis and by creating a fraction known as the waist-hip ratio. Values greater than 0.72 (primarily android fat distribution) are considered abnormal, but increased morbidity and mortality becomes more noticeable at values greater than 1.0.⁵⁴ The android pattern of obesity has been associated with abnormal lipid profiles and increased insulin resistance compared to the gynecoid pattern and has been attributed to different metabolic activity of fat according to site of deposition.

Acanthosis Nigricans

Acanthosis nigricans is a dermatologic condition marked by velvety, mossy, verrucous, hyperpigmented skin. It has been noted on the back of the neck, in the axillae, underneath the breasts and even on the vulva.⁵⁵ Originally it was noted in severe states of insulin resistance, as in the syndrome of hyperandrogenism, insulin resistance, and acanthosis nigricans (HAIR-AN syndrome). Like polycystic ovaries, however, its presence appears to be more a sign of insulin resistance than a distinct disease unto itself.⁵⁶ Like polycystic ovaries, there are other pathologic conditions associated with it that must be considered when noted. It also has appeared in association with malignant disease, especially adenocarcinoma of the stomach.⁵⁷

Hypertension

Hypertensive subjects are more likely to be hyperinsulinemic than controls who are normotensive. Insulin resistance persists in these subjects despite successful antihypertensive medication.⁵⁸ However, hypertension has not been consistently associated with PCOS, at least not in women of reproductive age.⁵⁹ This is a significant factor that prevents lumping all women with PCOS together under the diagnosis of insulin resistance syndrome. Long-term follow-up of women with PCOS may reveal an increased incidence of hypertension compared to age- and weight-matched controls.

Dyslipidemias

Many women with PCOS have significant dyslipidemias. Studies have shown that these women have lower HDL and/or HDL₂ levels and higher triglyceride and low-density lipoprotein (LDL) levels than age-, sex- and weight-matched controls.^60,61 Epidemiologic studies have shown that hyperlipidemia, including hypertriglyceridemia and elevated LDL/HDL ratios, is associated with an increased risk of atherogenesis. These associations have been found to exist independent of body weight in women with PCOS.⁶² These abnormalities are similar to those initially reported in syndrome X and are believed to reflect insulin resistance in these patients, rather than hyperandrogenism. No improvement in lipids was noted after suppression of ovarian hyperandrogenism with a gonadotropin-releasing hormone (GnRH) agonist.⁶³

Menstrual Irregularity

Because peripheral hyperandrogenic disorders such as hirsutism and acne are not invariably present in patients with elevated circulating androgens, menstrual irregularity may be the most clinically evident feature of PCOS. Menstrual irregularity and hyperandrogenism frequently exist independent of one another. However, when they are combined and produce hyperandrogenic chronic anovulation or PCOS, insulin resistance can almost always be detected. Menstrual irregularity is not, however, a specific marker of insulin resistance. It is frequently encountered in states of hyperandrogenic excess, such as congenital adrenal hyperplasia due to 21-hydroxylase deficiency, where insulin resistance has rarely been demonstrated.

Peripheral Disorders

As alluded to earlier, there is no evidence that insulin resistance as opposed to hyperandrogenism is involved in hirsutism. There has been at least one article suggesting that women with alopecia are hyperinsulinemic,⁶⁴ but there was also a strong correlation between testosterone and insulin levels. Unfortunately there have been limited studies of alopecia in women. The role of insulin resistance in other abnormalities of skin appendages noted in women with PCOS, such as acne vulgaris, is also unknown.

Cancer

Chronic anovulation places women with PCOS at increased risk for endometrial hyperplasia and endometrial cancer because of the long-term exposure to unopposed estrogen.^65,66,67 The association between PCOS and breast cancer is less well established, but case control studies have found elevated circulating androgens in both premenopausal⁶⁸ and postmenopausal women with breast cancer.⁶⁹ There is also evidence that hyperinsulinemia and insulin resistance, common features of PCOS, increase breast cancer risk.⁷⁰

Cardiovascular Disease

The evidence discussed thus far has suggested that women with PCOS, with known cardiac risk factors such as dyslipidemia and obesity, would be at increased risk for cardiovascular disease. Epidemiologic data, however, have not consistently identified hyperinsulinemia as an an independent risk factor for cardiovascular disease in women.⁷¹ A retrospective study of women with PCOS who had undergone wedge resection of the ovary 22 to 31 years previously, reported a fourfold higher prevalence of hypertension compared to age-matched controls⁷² and they had a relative risk for a myocardial infarction of 7.4-fold higher compared to controls.⁷³ A large case control study of women with PCOS found that they have significantly increased cardiovascular risk factors compared to controls, including elevations in body mass index; waist-hip ratio; insulin, cholesterol, LDL, and triglyceride levels; and systolic blood pressure. These women also had decreased HDL levels.⁷⁴

Unfortunately there are no long-term prospective studies of women with PCOS to document an increased risk of cardiovascular disease. It is important to note that in these retrospective studies the diagnosis of PCOS was made according to varying criteria, often based solely on menstrual history and/or ovarian morphology (i.e., wedge resection), and strict clinical and biochemical identification of these patients was not utilized (or not practical at the time of their initial presentation). Findings of long-term sequelae in an ethnically homogenous group (i.e., Scandanavian population) may not be applicable to the ethnically diverse population in America.

Diabetes Mellitus

The long-term follow-up of PCOS women from Finland showed a fivefold increased risk of diabetes mellitus compared to age-matched controls. Up to 20% of obese women with PCOS may have impaired glucose tolerance, which is substantially above the prevalence rates for glucose intolerance reported in population-based studies in women of this age (5.3% by National Diabetes Data Group criteria in women aged 20 to 44⁷⁵). NIDDM appears to result from a progressive disorder of glucose metabolism that is both preceded and predicted by insulin resistance.⁷⁶ These data suggest that PCOS is a major risk factor for NIDDM in women.

A suggested evaluation of women with PCOS for insulin resistance is found in Table 4. Unfortunately there is no accurate, inexpensive, reproducible test for assessing insulin sensitivity clinically. The gold-standard tests for assessing insulin sensitivity are the euglycemic clamp⁷⁷ and the frequently sampled intravenous glucose tolerance test.⁷⁸ Although these methods are the most accurate for determining insulin resistance, they are too laborious and expensive to use for a routine clinical evaluation of insulin resistance. Some studies have suggested some utility to a fasting glucose-insulin ratio in women with PCOS, although insulin resistance was only poorly assessed by the sum of serum insulin on an oral glucose tolerance test.⁷⁹ Insulin levels, either fasting or postprandial, have only a moderate correlation with more detailed studies of insulin action, and are not useful in patients with impaired glucose tolerance or frank diabetes.⁸⁰ Most important, there is a wide range of insulin sensitivity in normal populations and a substantial overlap of even the most precise measure of insulin action between insulin resistant and control populations. On an individual basis, therefore, it is often impossible to diagnose insulin resistance. Given the high prevalence of abnormalities in glucose tolerance, women with PCOS should be screened for glucose intolerance with an oral glucose tolerance test. A single random fasting insulin level provides less information. It is reasonable to consider all women with PCOS to be insulin resistant.

TABLE 4. Suggested Evaluation of Insulin Resistance in PCOS Women*

  Physical Examination
  Central obesity (hip-waist ratio)
  Acanthosis nigricans
  Alopecia (?)

  Laboratory

  Serum lipids

  Total cholesterol, HDL and LDL cholesterol, triglycerides

  Glucose tolerance

  75 g oral glucose tolerance test

*This assumes that other causes of hyperandrogenic chronic anovulation have been excluded.
HDL = high-density lipoprotein; LDL = low-density lipo-protein.

Therapeutic interventions are directed at specific symptoms of PCOS. Therefore, the goal of treatment may be control of hyperandrogenism, restoration of menstrual cyclicity, and ovulation induction to achieve pregnancy. Because these symptoms are not always attributed to insulin resistance, many traditional treatments have either no impact or a deleterious impact on glucose homeostasis. Interventions now exist to primarily increase insulin sensitivity, and limited data are available that examine the impact of other treatment goals on insulin sensitivity.

Weight Loss

Weight loss in obese women with PCOS has consistently been found to improve the reproductive abnormalities associated with the syndrome.⁸¹ Weight loss frequently improves menstrual cyclicity and decreases circulating androgen levels.⁸² In addition, weight loss decreases serum insulin levels.^83,84,85 Although studies have shown that exercise does not appear to offer any short-term beneficial changes in circulating insulin in women with PCOS,⁸⁶ there have been no studies of the long-term effect of exercise. Weight loss, through both dietary interventions and increased exercise, should be a cornerstone of treatment intervention in obese women with PCOS.

Pharmacologic Interventions

Older pharmacologic formulations that decrease insulin levels, such as diazoxide⁸⁷ or somatostatin,⁸⁸ also decrease serum androgen levels in women with PCOS. These agents are not useful clinically, however, because they worsen glucose tolerance or are poorly tolerated. Many newer pharmacologic agents that improve insulin sensitivity have been developed and are being studied. Such a possibility may also explain the effect of insulin-sensitizing agents, such as metformin and troglitazone, in women with PCOS. They result in decreased gonadotropins and circulating androgens. Insulin sensitivity is also improved, but this may be secondary to weight reduction. Metformin administration to women with PCOS has resulted in decreased hyperandrogenism and restoration of menstrual cyclicity.⁸⁹ Spontaneous pregnancies in patients on this agent alone have also been reported. The initial experience with troglitazone suggests that it also will significantly decrease insulin resistance as well as improve hyperandrogenism.⁹⁰ Further study of both the long-term sequelae of PCOS and interventions to improve insulin sensitivity in women with PCOS are needed before this treatment can be uniformly recommended.

Other Hormonal Interventions

Many women with PCOS are treated with oral contraceptives to restore menstrual cyclicity, reduce the exposure to unopposed estrogen, and suppress androgens. Because oral contraceptives may have adverse effects on insulin action, there is a theoretic concern about their use in women with PCOS with confirmed insulin resistance. A worsening of insulin sensitivity in women with PCOS was noted in a small study using a triphasic contraceptive.⁹¹ This is an area in which further investigation is needed to guide treatment strategy. In women in whom the primary concern is to achieve menstrual cyclicity to decrease the risk of endometrial hyperplasia or cancer, intermittent medroxyprogesterone acetate administration may be the treatment of choice because of its minor impact on insulin sensitivity.⁹²

The long-term use of glucocorticoids, because they can worsen insulin sensitivity, should be viewed with caution in insulin-resistant women with PCOS. Other additional interventions in women with hyperandrogenism include the use of androgen agonists and GnRH agonists. Spironolactone, even at high doses, does not impair glucose tolerance,⁹³ and may actually have some beneficial effects on insulin action.⁹⁴ However, it is frequently given in conjunction with oral contraceptives. Flutamide has also not been shown to affect insulin sensitivity in women with PCOS.⁹⁵ Antiandrogens must be used with caution in PCOS women of reproductive age, given their potential teratogenic effects in a male fetus.

Long-acting GnRH analogs can effectively suppress ovarian hyperandrogenism in insulin-resistant PCOS women with no significant effect on insulin resistance. This also produces a hypoestrogenic state that may result in sequelae such as osteoporosis and cardiovascular disease over a prolonged period of time. It is possible to add back low-dose estrogen replacement to prevent these events. While the use of the GnRh agonist alone in these women has not been found to alter insulin sensitivity, add-back hormone replacement therapy may have an effect. A preliminary study has suggested that low-dose add-back estrogen replacement alone does not lower insulin sensitivity (although adding a progestin did).⁹⁶ This also represents an exciting field for further research.

Hyperinsulinemia combined with hyperandrogenism in women has many potential etiologies. Unfortunately it seems we tend to understand the specific defect best in the rarest syndromes. PCOS is the most common cause despite being a diagnosis of exclusion. But this unique syndrome offers the tantalizing possibility of understanding the mechanisms that cause hyperinsulinemic hyperandrogenism. A substantial portion of the population would be benefited by such insight. The future of physician interaction with PCOS belongs not so much to the suppression of hyperandrogenism and ovulation induction, for which adequate diagnostic and therapeutic options exist, but to the diagnosis and treatment of insulin resistance, which comparatively is in its infancy. As the obstetrician-gynecologist expands his or her talents into the field of primary care, an expanded awareness of the signs, symptoms, and biochemical abnormalities of women with PCOS will help not only in the identification and follow-up of the long-term sequelae of hyperinsulinemic hyperandrogenism, but also in determining when and how to intervene.

Supported by grant HD01118-01 from the National Institute of Child Health and Human Development.