Assessing and treating insulin resistance in women with polycystic ovarian syndrome

Abstract

Polycystic ovarian syndrome (PCOS) is a highly prevalent hormonal and metabolic disorder among reproductive aged women worldwide. Women with PCOS have widely varying phenotypes and seek medical care for differing reasons. In addition to concern for menstrual cycle function, ovulation, hirsutism and acne, many PCOS women have abnormal glucose metabolism. While diabetes mellitus and impaired glucose tolerance are easily diagnosed, the diagnosis of and concern for insulin resistance as a precursor disorder is underappreciated. Insulin resistance may be the first important marker of metabolic disease in PCOS women at risk for metabolic syndrome and coronary artery disease.

Key Words: Polycystic ovarian syndrome; Insulin resistance; Impaired glucose tolerance; Diabetes mellitus; Infertility

INTRODUCTION

Diagnosis of polycystic ovarian syndrome (PCOS) is relatively straightforward. Common criteria established by the Rotterdam Conference in 2003 include at least two of three characteristics (oligomenorrhea, clinical and/or biochemical hyperandrogenism and ultrasound criteria) in the absence of other disease. PCOS is the most common hormonal disorder in women worldwide with prevalence estimates between 4%-8% but as high as 25% in some populations[1]. Women often initiate medical care for a cluster of PCOS symptoms (infertility, hirsutism and irregular menstrual cycles) that ultimately are not the most concerning medical consequences of PCOS [diabetes mellitus (DM), coronary artery disease (CAD), endometrial hyperplasia/cancer]. Here exists an important paradigm in the recognition and treatment of PCOS.

Clinically speaking, the hyperandrogenism seen in PCOS is associated with hirsutism more than acne or alopecia and therefore hirsutism is an impetus for young women seeking care[2]. Many PCOS women are also overweight (BMI > 25kg/m²) or obese (BMI > 30kg/m²), although adiposity is not a defining criteria for PCOS. Obesity is highly prevalent in the general population and in PCOS women and is an independent risk factor for CAD[3]. Obesity in adolescents is correlated with insulin resistance (IR) and dyslipidemia[4]. PCOS related ovulatory dysfunction in adolescents often correlates to adolescent obesity[5]. Genetic predisposition to PCOS has been suspected for many years[6] and data link obesity and metabolic disturbances in PCOS with genetic polymorphisms[7,8]. Even male first degree relatives of women with PCOS have a higher incidence of metabolic syndrome (MS), the closest corollary to PCOS in men[9].

Once a diagnosis of PCOS is confirmed, it is imperative to assess women for CAD risk factors. Despite the many reasons women seek medical care for PCOS, the greatest long term risk for these women is CAD. This is generally not viewed or even recognized as a concern by women seeking care in the first place. The link between PCOS and CAD is multi-faceted. C-reactive protein (CRP) is higher in age matched PCOS women and is linked to BMI[10] with some ethnic variation in this risk[11]. The prevalence of MS in PCOS women is as high as 40% with increased prevalence of hypertension, dyslipidemia and abnormal glucose metabolism, all before age 30[12]. PCOS women aged 20-40 already demonstrate poor vascular function measured by brachial artery vascular flow[13]. No single blood test can predict or quantify this CAD risk. Although no standard recommendation for assessment of CAD risk factors exists, measurement of glucose metabolism, blood pressure screening, lipid screening and carotid intimal media thickness measurements have been suggested[14].

The routine use of OGTT is advocated by some in all PCOS women[15]. In teenagers, abnormalities in glucose metabolism manifest prior to dyslipidemia, suggesting that assessment of glucose metabolism is even more important in younger women[16]. DM is diagnosed by an 8 h fasting plasma glucose ≥ 126 mg/dL, 2 h glucose value ≥ 200 mg/dL after oral glucose tolerance test (OGTT) or random glucose ≥ 200 mg/dL with symptoms of DM confirmed by either fasting plasma glucose or OGTT. Hemoglobin AIC > 6.5% may also be issued to diagnose DM[17]. Impaired glucose tolerance (IGT) is defined by a 2 h cutoff of 140-200 mg/dL on OGTT[18].The prevalence of IGT in obese adolescents is surprisingly as high as 15%[19].

INSULIN RESISTANCE AND PCOS

As many as 70% of PCOS women are insulin resistant and 10% have DM[20-22]. In PCOS women with normal glucose metabolism initially, the rate of conversion to abnormal glucose metabolism can be 25% over just three years[23]. More alarming, insulin abnormalities are highly prevalent in adolescents with PCOS[24]. Almost 20% of young Thai women with PCOS actually have DM[25]. Overall, normal glucose levels on an OGTT do not predict IR and IR, despite normal glucose levels, is correlated with CRP, dyslipidemia and other CAD risk factors[26]. Therefore, glucose levels alone lack the sensitivity to predict metabolic risk in PCOS patients. Precursor states of insulin abnormalities likely predict long term CAD risk well before glucose abnormalities. IR can be just as severe in diabetics and non-diabetics[27], stressing the seriousness of this metabolic impairment as a precursor and not a separate disease. Animal models have shown that IR alone damages myocardial cells, providing direct evidence of end organ disease[28]. Human data link HOMA-IR to left ventricular dysfunction[29]. Abnormal glucose metabolism short of IGT and DM still deserves attention, identification and treatment[30].

PCOS women with different phenotypes have been found similarly insulin resistant in response to a 3 h 75 g OGTT[31]. Obese (compared to lean) PCOS women tend to have a higher degree of IR. Correlation between hyperandrogenism and IR is significant in many studies but not as significant as the link between insulin abnormalities and obesity[32]. PCOS women demonstrate greater variation in insulin parameters compared to controls, independent of weight[33]. Animal studies of prenatal testosterone exposure show downstream IR in early postnatal life[34]. Some human data shows a high degree of correlation between hyperandrogenism and IR[35,36] and the relationship between hyperandrogenism and IR seem to differ between PCOS and non-PCOS women[35].

Reproductive dysfunction in PCOS women may also be a manifestation of IR. Menstrual cycle irregularity has been correlated with HOMA-IR[37]. Molecular defects in insulin action may be responsible for reproductive difficulties in PCOS women. Although endometrial tissue appears morphologically similar in PCOS to controls and may have similar insulin receptor prevalence, insulin receptor action at the local endometrial level is impaired and may be reflected in lower pregnancy implantation rates[38]. HOMA-IR has been correlated with follicle count in PCOS during in vitro fertilization[39]. Follicular insulin levels correlate with pregnancy outcome after IVF[40]. These are areas of unresolved understanding with regard to PCOS. Proposed mechanisms for insulin reproductive abnormalities include abnormalities of ovarian steroidogenesis, excessive LH secretion and abnormalities in glucose uptake[41]. PCOS women have been found to have post-receptor insulin abnormalities as well as reduced peripheral insulin receptor binding[42].

ASSESSMENT OF INSULIN RESISTANCE

No universal definition of insulin resistance exists and therefore no standard clinical technique to measure insulin resistance exists. Insulin resistance can be thought of as a metabolic state where normal glucose homeostasis mechanisms fail to operate properly. Translating theory to clinical practice has been a source of frustration for many practitioners. The American Diabetes Association has characterized IR as a state of impaired metabolic response to insulin[43]. IR is characterized by an inability of normal amounts of insulin to achieve the normal predicted response, often in the clinical setting of central adiposity. To achieve euglycemia, the pancreas over secretes insulin[44]. Investigators define IR based on hyperinsulinemic-euglycemic clamp techniques as a state of impaired glucose disposal in response to insulin[22]. Despite no consensus, clamp techniques have become the reference for understanding IR.

Hyperinsulinemic-euglycemic clamp techniques rely on an intravenous insulin infusion to maintain steady serum glucose concentrations at fasting levels to measure glucose uptake. Lower glucose uptake signifies resistance to insulin action (i.e. IR). Since the technique requires intravenous infusions, frequent blood sampling, extensive time and significant financial resources, it is experimentally useful but clinically cumbersome[45]. Clamp studies in PCOS women show conflicting results; some studies show IR only in obese PCOS women[46] and others demonstrate IR in lean PCOS patients[47]. Of importance, the studies which failed to demonstrate IR in lean PCOS women did, however, demonstrate elevated basal insulin levels compared to weight matched, non PCOS controls[46]. Other sophisticated testing methods using intravenous infusions of insulin have been attempted (insulin sensitivity test and insulin tolerance test) but they do not alleviate the time, financial and testing burdens to make them relevant for widespread clinical practice and normal cutoffs are not widely disseminated[45]. Clamp techniques have been used as comparisons to validate other modes of assessment of IR.

Fasting methods to measure IR have been advocated for many years as an adjunct to DM screening. Elevated fasting insulin levels greater than 20 μU/mL may alone indicate IR. Fasting glucose/insulin ratio (G/I) has also gained some clinical traction. A ratio < 4.5 has in general been shown to be > 90% sensitive in some populations[45] but has never been validated with clamp studies[48]. Some ethnic variation in G/I cutoff ratios may exist[49]. There has been some suggestion that G/I < 7 in very young girls may predict IR[50,51].

The homeostatic model assessment (HOMA), a more complex fasting calculation, has been compared to clamp techniques with good results. HOMA is the product of fasting glucose (mg/dL) and insulin (μU/mL) divided by a constant[45]. One major limitation of HOMA rests on the previous reflection that many young PCOS women display stimulated but not fasting metabolic abnormalities. In fact, HOMA in young PCOS patients missed 50% of IR as compared to OGTT with insulin-AUC calculations[52]. G/I ratio correlated strongly with clamp-demonstrated IR in a small study of PCOS women - interestingly, both lean and obese PCOS women had evidence of IR. Sex hormone binding globulin (SHBG) did not correlate with IR in this study[47], as has been previously postulated[53].

Quantitative insulin sensitivity check index (QUICKI) was developed to improve the sensitivity of fasting measurements. QUICKI is calculated as: 1/[log(insulin fasting) + log(glucose fasting)] and has been well correlated to clamp measurements in obese and non-obese patients[15]. QUICKI also demonstrates correlation with HOMA-IR[53]. QUICKI research calculations in young PCOS women are often identical to age matched women with DM[54].

OGTT with 75-g glucose and hourly glucose and insulin measurements has been compared to clamp techniques. Insulin sensitivity calculated by mathematical transformation of measurements has shown good correlation with glucose disposal using clamp techniques[48]. Although the OGTT is easy to perform, these calculations are more complex and make this particular calculation less desirable for clinical use. However these data show that 1 and 2 h levels are often needed to diagnose IR and stress the potential for false negative results with fasting measurements alone. In patients undergoing clamp and OGTT no correlation was observed between fasting glucose/insulin ratios and IR on the clamp[48].

Some have tried to utilize ultrasound to detect IR. Of note, normoglycemic women often have the phenotypic criteria for polycystic ovaries on ultrasound[54], consistent with other data in young adolescents showing that polycystic ovaries by ultrasound appearance often does not correlate with either anovulatory menstrual cycles or metabolic abnormalities[55]. Therefore ultrasound is too non-specific to use with any reliability in measuring IR.

Limitations of direct insulin testing and cumbersome calculations have led to research for indirect serum markers to provide evidence of IR. SHBG correlations to IR as previously mentioned have been inconsistent. Adiponectin is a protein found in adipose tissue associated with both inflammation and insulin action. Recent studies have linked plasma adiponectin level to IR (but not hyperandrogenism) measured by HOMA[56-58]. Serum soluble glycoprotein-130 levels (local cytokine) have been inversely correlated to IR[59]. Resistin plasma levels have been correlated with fasting glucose and HOMA-IR in PCOS women[60]. Inhibin A levels in PCOS women were not found to correlate with IR in PCOS women[61]. Most of these serum markers share common limitations and have been poorly studied. How they might vary with different PCOS phenotypes is unknown. None are adequately compared to IR measured by clamp studies. Their usefulness serially in clinical practice to monitor patients over time and undergoing treatment is also unknown. Some genetic work has recently shown promise. Although far from clinical use, microarray analysis of genes in muscle, adipose tissue and the liver shows alterations in the setting of IR[62]. Serum genetic markers may lead to future genetic techniques to detect and monitor IR.

TREATMENT

Why treat IR in PCOS women? For many years only PCOS women with DM were treated. As the link between IGT and CAD became more apparent, many PCOS women with IGT were treated. We now understand that IR is often the first step in a progression to DM and CAD. Those who now advocate treatment for IR do so for the following reasons: reduction of insulin and androgen levels, prevention of IGT and DM, potential for improved ovulation, symptomatic improvement, prevention of MS[63]. Ultimately, secondary prevention in young women with identifiable and treatment precursor conditions is far more desirable and easier than treatment of these same women later in life with serious disease.

Metformin has been the mainstay of treatment for IR and IGT in PCOS women over the past decade. Metformin is a biguanide that acts principally on the liver to inhibit hepatic gluconeogenesis. It also inhibits acetyl-CoA carboxylase activity and suppresses fatty acid production. Metformin acts on skeletal muscle to inhibit lipid production and acts peripherally on adipose tissue to stimulate glucose transport and uptake. Metformin reduces insulin levels and promotes improved insulin receptor activity[64]. Metformin may also have direct and indirect effects on the ovary with respect to insulin action and steroidogenic enzymatic activity. In the endothelium, metformin seems to improve nitric oxide vasodilatory effects. Many other mechanisms of action have been studied in both animal and human models but consistent effects are not always demonstrated with local tissue concentrations that result from therapeutic doses[65].

Human data regarding metformin improvement in IR in PCOS women shows mixed results and is complicated by varying methods of assessing IR. Short term (3 mo) treatment with metformin (1500 mg per day) failed to affect IR as measured by AUC-Insulin after 75-g OGTT. Metformin (1600 mg per day) in obese PCOS women treated for 6 mo failed to reduce IR as measured by QUICKI[66]. This is in contrast to similar length studies on obese PCOS women who demonstrated decreased IR as measured by HOMA-IR, QUICKI and ISI, and correlated with alterations in phosphoproteins related to IR[67]. Longer term metformin therapy (2 years, 1600 mg per day) in young, obese PCOS women reduced fasting insulin, hyperandrogenism and produced borderline reductions in HOMA-IR (P = 0.05)[68]. Metformin was compared prospectively to naltrexone and prenisolone in combination with oral contraceptive pills (OCPS). IR was unchanged despite lowered androgen levels[69]. Metformin has been compared to orlistat and pioglitazone over a 4 mo treatment course and although each treatment reduced IR as measured by HOMA-IR, metformin (1500 mg per day) had the least reduction (< 20%)[70].

Studies have attempted for years to show an advantage to metformin for ovulation induction and as an adjunct to more advanced fertility treatments. In ovulatory PCOS women metformin was associated with improved serum and follicular fluid AMH levels as well as insulin values; these changes were not seen in anovulatory PCOS women[71]. Despite the demonstration of negative effects of IR on reproductive outcome, the vast majority of evidence does not show improvement in live birth rates when metformin is used strictly for fertility[72], although treatment does improve ovulatory status[72,73].

Metformin has been studied specifically in adolescent PCOS women. Metformin therapy for 10 mo decreased fasting serum insulin levels in obese girls with PCOS[74].The positive effects of metformin in adolescents wore off within 3 mo of medication discontinuation[75]. Metformin in obese PCOS adolescents has shown improvements in IR by clamp studies, fasting measurements and OGTT after just 3 mo of therapy[76,77]. Other studies have found non-significant trends to improved IR by HOMA and OGTT-AUC in adolescent PCOS patients[78]. Metformin has also been shown to effectively contribute to BMI reduction in PCOS adolescents[79].

Metformin has been tested in combination with cholesterol lowering medications. Pretreatment of obese PCOS patients with atorvastatin (20 mg per day for 3 mo) followed by 3 mo of metformin (1500 mg per day) resulted in more effective lowering of HOMA-IR than metformin alone[80]. Other similar data show that combined treatment with metformin and atorvastatin compared to metformin alone produced similar but significant improvements in IR. Combination therapy only showed successful reduction of hyperandrogenism and not IR[81].

The ultimate goal is to prevent metabolic disease. Metformin (1500 mg per day) compared to placebo in a prospective 12 wk randomized control trial decreased arterial stiffness (by peripheral pressure waveforms in the brachial artery) and endothelial function (measured by augmentation index). Metformin did not reduce HOMA-IR[82]. The study population was obese but young (mean age 30 years), demonstrating the ability to reduce CAD risk even in very young women. Metformin has reduced both carotid intimal media thickness and endothelin levels in obese PCOS women[83]. In many studies metformin has reduced both total cholesterol and LDL cholesterol levels[84-86], triglyceride levels[84] and increased HDL levels[87,88]. Animal studies have shown that acarbose given to insulin resistant rats decreased carotid intimal hyperplasia and blood flow velocities[89]. Taken as a whole, the ability of metformin (and likely other insulin sensitizing agents) to elicit an overall reduction in the risk for CAD may be easier than the ability to produce consistent measureable improvements.

Other insulin sensitizing agents have been advocated and studied for the treatment of IR in PCOS, principally thiazolinediones. Thiazolinediones stimulate gene transcription that alters lipid and glucose metabolism, decreases lipolysis and decreases fat deposition[90]. Thiazolinediones decrease fatty acid release, suppress gluconeogenesis and reduce tumor necrosis factor α disruption of insulin activity[64]. Pioglitazone and rosiglitazone have decreased IR (measured by clamp studies) in PCOS women[90-93]. Glitazones have also decreased IR by OGTT AUC-Insulin in PCOS women[91,93,94]. In patients with DM, thiazolinediones reduce central adiposity[95], a trait commonly shared with PCOS women. Pioglitazone by way of IR and adiponectin levels also has improved menstrual regularity in PCOS women[96,97]. Adverse outcomes have been seen in pregnant animals with limited to no human data. Therefore, as a class, thiazolinediones are not considered first line therapy for PCOS women seeking pregnancy. Rosiglitazone has even been found to decrease pro-inflammatory markers in human granulosa cells cultured following in vitro fertilization oocyte retrieval, thus showing additional target tissue for therapy[98]. However, these effects have not been adequately studied and have no current practical application.

Other pharmacological treatments have attempted to lower IR. Vitamin D has been shown to decrease HOMA-IR despite a lack of change in hyperandrogenism in young, obese PCOS women[99]. Animal studies have demonstrated that treatment with glycyrrhizic acid affecting lipoprotein lipase activity decreases serum insulin and HOMA-IR[100]. Although oral contraceptive pills positively affect hyperandrogenism, they have little to no effect on glucose metabolism by OGTT[101]. Long term oral contraceptive pill use may have some limited benefit in IR but data are limited[102]. A 6 mo course of oral contraceptive pill treatment in adolescent obese PCOS women has demonstrated some improvement in IR[103].

Lifestyle interventions are usually required for long term sustainable results. PCOS women who smoke have higher free androgen levels and IR as measured by HOMA-IR, QUICKI and the insulin sensitivity index following 75 g OGTT[104]. Thus PCOS women who smoke have an additional reason to stop smoking. In more general population studies (non-PCOS) comprised mostly of middle-aged women, lifestyle intervention is more effective than metformin in preventing the progression to DM. Dietary and exercise intervention decreased the 4 year progression to DM in patients at risk (non-diabetic, elevated fasting and/or OGTT glucose) by almost 50%[105]. Realizing the limitations of applying this population sample to young PCOS women, it still highlights the benefit of non-pharmacological treatment. PCOS women randomized to both metformin and lifestyle interventions (compared to placebo) showed improvements in HOMAIR after 4 mo[106]. In European adolescents with PCOS who failed to achieve improvements in HOMA-IR after 6 mo of lifestyle intervention, both metformin and placebo reduced IR over 6 mo, although metformin offered no benefit over placebo[107]. Lifestyle modification in adolescents has been successful in reducing hyperandrogenism[103]. Modest weight loss of about 5% bodyweight has also been shown to lower hyperandrogenism[108] which may ultimately improve IR.

Acupuncture has been studied as a means to reduce IR in PCOS phenotype animals. Acupuncture decreased IR by euglycemic-hyperinsulinemic clamp and altered glucose transporter expression (GLUT4) in a rat model of PCOS[109]. In humans, acupuncture has shown both metabolic and hormonal benefits in women with PCOS[110].

CONCLUSION

Regardless of what reasons women have for seeking diagnosis and treatment of PCOS, it is imperative for practitioners to assess a woman’s risk for CAD. Assessment should probably be made in all PCOS patients regardless of BMI. Especially in young women or adolescents, IR may be the first identifiable risk factor. Practitioners must recognize that no universal test for IR exists and must use good clinical judgment to assess metabolic status in women. Stimulated testing with OGTT may be more sensitive than fasting measurements. Women who demonstrate IR should be counseled on lifestyle modifications. Physicians should discuss with their patients a target BMI that is realistically obtainable. It is often advisable for patients to seek nutritional assessment and counseling to help with this goal. In many individuals, consideration should be given to pharmacological treatment. Although the most commonly used medication is metformin, other medications may be appropriate first line therapy, especially in women not actively seeking pregnancy.