Surrogate end-points or primary outcomes in clinical trials in women with polycystic ovary syndrome?

Richard S. Legro1,3 and Evan Myers2 for the Reproductive Medicine Network

1 Department of Obstetrics and Gynecology, Pennsylvania State University College of Medicine, Hershey, PA and 2 Department of Obstetrics and Gynecology, Duke University, Durham, NC, USA

3 To whom correspondence should be addressed at: Department of Obstetrics and Gynecology, PO Box 850, 500 University Drive, M.S. Hershey Medical Center, Hershey, PA 17033, USA. Email: rsl1{at}psu.edu


    Abstract
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
There are multiple surrogate variables in polycystic ovary syndrome (PCOS), including biometric and biochemical parameters. The number of surrogate variables and their poor validity in relationship to primary clinical end-points pose major problems to conducting a trial in women with PCOS. The aim of this review is to discuss the use of surrogate variables compared with primary clinical end-points in women with PCOS. Arguably the best documented correlation between a surrogate variable and a primary clinical end-point is that between ovulation and pregnancy in women with PCOS. Good correlation has been noted between the increase in ovulation frequency with clomiphene citrate and the chance of pregnancy in women with PCOS. However, ovulation cannot be equated with pregnancy, as a host of other factors may affect the true outcome of interest: a healthy liveborn child. Pregnancy and an improvement in hirsutism are clinical end-points that have been successfully studied in past and ongoing clinical trials in women with PCOS. Many other clinical end-points, such as endometrial cancer and cardiovascular disease, are rare in premenopausal women with PCOS, and may not be suitable as the primary outcome of clinical studies. Future multicentre trials in women with PCOS should focus on primary clinical end-points.

Key words: diabetes prevention/hirsutism/hyperandrogenism/insulin resistance/pregnancy/randomized clinical trial/surrogate measure


    Introduction
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
Polycystic ovary syndrome (PCOS) is one of the most common endocrinopathies among women, with epidemiological studies of its prevalence suggesting that ~5% of the female population in the Western world is affected with irregular menstrual periods and androgen excess (Knochenhauer et al., 1998Go; Diamanti-Kandarakis et al., 1999Go; Asuncion et al., 2000Go). Despite this high prevalence, there are only limited data from randomized controlled trials which demonstrate that any currently available therapy for this condition results in improvement of primary medical end-points such as infertility, hirsutism and endometrial cancer. Simply put, primary end-points are what is of interest to patients, and surrogate variables are generally of interest to clinical researchers. While this is a general problem for many issues in women's health, this paper focuses on potential reasons for this lack of evidence guiding the treatment of women with PCOS, and suggests new avenues for clinical research in the syndrome.

PCOS is a heterogeneous disorder, one that has evolved over the years from a ‘disease’ to a ‘syndrome’ (note in the last decade the disappearance of the acronym ‘PCOD’ and the phrase ‘polycystic ovarian disease’ from the published literature). A disease suggests a more fully characterized set of symptoms than a syndrome, which is a looser association of symptoms. PCOS is a heterogeneous disorder, where current proposed diagnostic criteria include a number of disorders with similar phenotypes, but different aetiologies. However there is no single sign or symptom that is currently accepted as the sine qua non of the syndrome. Therefore, there is no gold standard test of this sign or symptom from which to determine the sensitivity/specificity of screening tests, entry criteria into a clinical trial or long-term sequelae.


    Potential diagnostic criteria in PCOS
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
There are a variety of diagnostic criteria that have been proposed for PCOS (Table I). Historically, the diagnosis as first described by Stein–Leventhal, involved enlarged ovaries, obesity, hirsutism and infertility due to chronic anovulation (Stein and Leventhal, 1935Go). As urinary assays for gonadotropins developed, an emphasis on inappropriate gonadotropin secretion emerged, with abnormalities in LH hypersecretion (McArthur et al., 1958Go). With the development of sensitive and specific assays for circulating androgens, hyperandrogenemia came to the forefront of the diagnostic criteria (Raj et al., 1978Go). As abdominal and then transvaginal ultrasonography reached ascendancy, emphasis was placed by some on the ovarian morphology as determined by ultrasound (Adams et al., 1986Go). As the role of insulin resistance and hyperinsulinemia has emerged most recently, clinical tests of insulin action have been also promoted as part of the diagnostic evaluation in the syndrome (Legro et al., 1998Go). The prevalence of the syndrome will always depend on the chosen diagnostic criteria, which are subject to local prevailing winds (Balen and Michelmore, 2002Go). For instance, it tends to be higher, in the range of 20–30%, when polycystic ovaries are used as the initial identification of subjects (Michelmore et al., 1999Go). Furthermore, ethnic and racial background also appear to influence the PCOS phenotype, and much higher prevalences (~50%) have been reported in such populations as South Asians from the Indian subcontinent (Rodin et al., 1998Go; Wijeyaratne et al., 2002Go).


View this table:
[in this window]
[in a new window]
 
Table I. Previous possible diagnostic criteria in PCOS: some of the historical criteria that have been used or suggested to make the diagnosis

 
The diagnostic criteria established at the 1990 National Institutes of Health–National Institute for Child and Human Development (NIH-NICHD) conference on PCOS (hyperandrogenism and chronic anovulation with the exclusion of secondary causes such as adult-onset congenital adrenal hyperplasia, hyperprolactinemia, thyroid disorders and androgen-secreting neoplasms) have been widely accepted and utilized in clinical trials (Zawadski and Dunaif, 1992Go). This approach of combination diagnostic criteria has been recommended by most authors as superior to any single diagnostic criterion (Fox et al., 1991Go; Norman et al., 1995Go; Koskinen et al., 1996Go). This definition of hyperandrogenic chronic anovulation has also been utilized for several large series describing metabolic sequelae in PCOS (Talbott et al., 1995Go; Legro et al., 1999Go; Talbott et al., 2000Go; Legro et al., 2001Go; Wild, 2002Go), several seminal clinical trials in PCOS (Nestler and Jakubowicz, 1996Go; Nestler et al., 1999Go; Azziz et al., 2001Go), as well as in familial and genetic studies (Urbanek et al., 1999Go; Kahsar-Miller et al., 2001Go; Bo et al., 2004Go). These varying trials, and especially multicentre trials, must have uniform criteria to assure that there is uniformity in patient selection between sites. Thus, there appears to be emerging consensus, from both the experts and those toiling in the trenches of clinical research, towards the use of unexplained hyperandrogenemic chronic anovulation as the central diagnostic criteria in PCOS. A recent jointly sponsored ASRM/ESHRE Expert Consenus Conference maintained this as part of the recommended diagnostic criteria and recommended also adding polycystic ovary morphology as a potential diagnostic criterion (Table II) (The Rotterdam ESHRE/ASRM Sponsored PCOS Consensus Workshop Group, 2004aGo; bGo). It is too soon to judge the impact of these revised criteria, although the recommendation to clearly define and state the diagnostic criteria in publications is a good start.


View this table:
[in this window]
[in a new window]
 
Table II. 1990 NIH-NICHD Diagnostic Criteria for PCOS along with the Revised 2003 Rotterdam Criteria

 

    Definition of surrogate end-points
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
The preceding discussion of diagnostic criteria in PCOS introduces several variables that have been used as surrogate measures of treatment response. A surrogate outcome in the context of clinical trials has been defined as ‘a laboratory measurement or a physical sign used as a substitute for a clinically meaningful end-point that measures directly how the patient feels, functions or survives. Changes induced by a therapy on a surrogate end-point are expected to reflect changes in a clinically meaningful end-point.’ (Temple, 1995) The benefits of choosing to study surrogate variables instead of primary outcomes in clinical trials are that they may shorten the period of study, lower the sample size required and lower the costs of the study. Ideally, if there is a strong correlation between change in the surrogate variable and the primary clinical end-point (i.e. cardiovascular event rates, cancer survival rates, pregnancy rates), the effectiveness of therapy on the primary end-point can be quantified by estimating the proportion of the treatment effect explained along with its standard error (Lin et al., 1997Go; Flandre and Saidi, 1999Go). This variable, however, contains more than a mere correlation, as it must also account for the rates of false-positives and false-negatives.


    Discussion of surrogate end-points
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
One of the most studied and validated surrogate variables has been the change in circulating lipid levels and the effects on cardiovascular event rates. Cholesterol lowering has significant benefit on preventing cardiovascular-related mortality (Expert Panel on Detection Evaluation and Treatment of High Blood Cholesterol in Adults, 2001Go). Specifically, it has been found that for every 10% reduction in circulating cholesterol levels, there is a 13% improvement in cardiovascular related mortality and a 10% reduction in total mortality (Gould et al., 1995Go). This surrogate end-point of cardiovascular disease is so well accepted that it can be used as a measure to obtain approval of a new drug or to expand the labelling of an approved drug by the US Food and Drug Administration (FDA).

However, even such a well-validated surrogate variable can be misleading. For instance, in the Heart and Estrogen/Progestin Replacement Study trial, although combined hormone replacement therapy with continuous conjugated equine estrogens and medroxyprogesterone acetate resulted in favourable changes in the circulating lipid and lipoprotein levels in women with pre-existing heart disease, it was also associated with an increased cardiovascular event rate in the first year of the trial and no benefit in overall mortality over the course of the trial (Hulley et al., 1998Go). Similarly, increased cardiovascular events were seen in the Women's Health Initiative study of primary prevention of cardiovascular disease, despite favourable changes in lipid profiles with lowered levels of LDL cholesterol and elevated levels of HDL cholesterol in the treated groups (Writing Group for the Women's Health Initiative Investigators, 2002Go; Manson et al., 2003Go).


    Critique of surrogate end-points
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
Although the exact mechanisms for the failure of combined hormone replacement therapy with continuous conjugated equine estrogens and medroxyprogesterone acetate to provide either primary or secondary cardioprevention in women is unknown, one of the proposed causes is the tendency towards thrombophilia that this combination may encourage (Cushman et al., 1999Go; Rosendaal et al., 2002Go). Thus, beneficial effects on one system (circulating lipids and lipoproteins) may be negated by adverse effects on other systems (clotting). In addition to the effects of a treatment intervention on another biological system, Holloway and Dick (2002)Go suggested that the closer the surrogate outcome is to the primary clinical end-point on the causal pathway, the better it is (Figure 1). The further away it is, the greater the likelihood that it can be misleading.



View larger version (25K):
[in this window]
[in a new window]
 
Figure 1. Surrogate outcomes. (A) The ideal surrogate is in the causal pathway to the clinically meaningful end-point and mediates the effect of the intervention on the clinically meaningful end-point. (B) Surrogates can fail for several reasons. First, the surrogate may be transient or may not be of sufficient size or duration to alter the clinically meaningful end-point (solid line). Secondly, there may exist other causal factors (disease-related or ‘noise’) that affect the clinically meaningful end-points not captured by the surrogate (dotted line). Finally, an intervention may have unintended consequences on the clinically meaningful end-point (dashed line). Adapted from Holloway and Dick (2002)Go.

 
Ideally, there is a sole causative pathway and the surrogate marker is close to the end of it. For example, although vaccines against oncogenic human papillomavirus (HPV) are ultimately aimed at preventing cervical cancer, the relative rarity of cervical cancer, the long latency time from initial HPV infection and the ethical requirement to follow patients and treat those with high-grade cervical intraepithelial neoplasia (CIN) preclude the use of cancer as the end-point in clinical trials. However, because most HPV infections and most early CIN lesions regress, use of these as surrogates might overestimate the effectiveness of the vaccine. Therefore, current guidance from the FDA is that trials of prophylactic vaccines against oncogenic types use CIN 2/3 as the primary end-point (Goldenthal, 2001Go).

In order to determine its location on the causal pathway, a full understanding of the aetiology of the disorder is necessary, as well as studies validating the relationship between the surrogate marker and the primary clinical outcome. A grading scheme of surrogate variables based on their validity has been proposed (Prentice, 1989Go). Given our incomplete understanding of the pathophysiology of PCOS, it is difficult to grade the location of any surrogate marker in its proximity to the unknown source of the disorder. Ideally, the best surrogate marker should predict the outcome of interest, both with and without the proposed intervention. Such a marker has to have been studied extensively with both natural history studies and randomized clinical trials, both of which are underrepresented in the PCOS literature. The consequences of using non-validated surrogate markers includes, in a best-case scenario, patient benefit, but in other scenarios they range from wasted resources and ambiguous results to frank patient harm (Holloway and Dick, 2002Go).

A classic example of this latter scenario is the use of suppressing ventricular ectopic beats as a surrogate for reducing cardiovascular-related mortality after myocardial infarction (Anonymous, 1989Go). Because natural history studies showed that ventricular arrhythmias after a myocardial infarction increased the risk of subsequent death, it was hypothesized that if post-myocardial infarction arrhythmias were suppressed, then there would be a greater chance for survival. Eventually two new drugs, encainide and flecainide, were approved by the FDA on the basis of suppressing post-myocardial infarction arrhythmias. Their effect on the primary clinical end-point survival after a myocardial infarction was then evaluated in the Cardiac Arrhythmia Suppression Trial (CAST). This trial was terminated early because of excessive mortality in the flecainide and encainide treatment arms compared with the placebo arm of the trial. A similar lack of benefit to the post-myocardial patient were found with moricizine (Anonymous, 1992Go) and lidocaine (Hine et al., 1989Go; Holloway and Dick, 2002Go).


    Surrogate end-points in PCOS
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
There is no lack of surrogate end-points in PCOS. The number of surrogate variables and their poor correlation with primary clinical end-points pose major problems to conducting a trial in women with PCOS. The number of surrogate variables includes all of the previously proposed possible diagnostic criteria found in Table I, as well as other related variables such as circulating lipids and lipoproteins and measures of obesity. Within these categories of surrogate variables there are several subvariables, for instance body weight, body mass index, waist–hip ratio, skin fold thickness, body composition on DXA scan, body fat distribution on single slice computed tomography scan, etc., all of which can be used as measures of obesity. As illustrated by the example of obesity, there are a number of possible surrogates within the larger category of the surrogate marker.

The subcategory of hyperandrogenemia within hyperandrogenism offers a number of potential circulating androgen markers, including testosterone, androstenedione, dihydrotestosterone, and even weak androgen precursors such as DHEA and DHEAS. There are no standardized commercial assays for measuring total levels of these hormones in the blood (Boots et al., 1998Go). Furthermore, because androgens circulate bound to various proteins and primarily to sex hormone-binding globulin (SHBG), there are a number of measures of circulating testosterone that attempt to measure the bioavailability of testosterone. These include assays for free testosterone usually by equilibrium dialysis, measures of the fraction of serum T not precipitated by 50% ammonium sulfate concentration (non-SHBG-T) (Tremblay and Dube, 1974Go), and measures of circulating testosterone corrected for levels of SHBG and other proteins (i.e. albumin) like the free androgen index (Imani et al., 2000Go). While these latter measures are felt to better reflect the clinical presentation than total testosterone levels (Vermeulen et al., 1999Go), they have not been well validated in the realm of clinical trials.

Insulin sensitivity is another surrogate measure that is frequently investigated in clinical trials of women with PCOS. The gold standard test for determining insulin sensitivity (or at least insulin-mediated glucose uptake) is the euglycemic clamp, but this test is invasive, expensive and unwieldy for use in clinical practice (DeFronzo et al., 1979Go). Surrogate markers of insulin-mediated glucose uptake sensitivity are even more varied, and can rely on fasting measures of insulin and glucose as well as changes over time during dynamic challenges (Matsuda and DeFronzo, 1999Go). In addition, a number of surrogate markers of the insulin resistance syndrome, such as plasminogen-activator inhibitor-1 (Velazquez et al., 1997Go), C-reactive protein (Kelly et al., 2001Go) and homocysteine (Yarali et al., 2001Go), have been examined in women with PCOS. However, there are no universally accepted or validated markers of insulin resistance that have been recommended for use in clinical practice (American Diabetic Association, 1998Go; Bloomgarden, 2003aGo; bGo).


    Critique of surrogate end-points in PCOS
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
There are multiple problems with the number of surrogate end-points utilized in clinical trials in women with PCOS. First, the examination of multiple variables can lead to type I errors, as there is likely to be a significant change in one of the variables studied over the course of a clinical trial perhaps due to chance alone. Such findings should be corrected by the number of analyses attempted. Post-hoc analysis may also lend itself to a focus on the variable with significant change as the point of the trial, rather than the ones that remained unchanged or changed unfavourably. When multiple variables are examined in a clinical trial, the primary outcome variable should be identified during the trial design and utilized to perform the power analysis prior to initiating the trial (Moher et al., 2001Go). Few of these surrogate variables have been well validated as predicting the primary clinical outcomes in women with PCOS. These primary outcomes should be the ultimate goals of treatment (Figure 2). Like the prisoners chained in the cave in Plato's ‘Allegory of the Cave’, we have mistaken the shadows on the wall (i.e. the surrogate markers) for the the dancers on the rampart (i.e. the disease itself), and it is time to turn and face the true outcomes of interest to our patients.



View larger version (170K):
[in this window]
[in a new window]
 
Figure 2. Illustration of Plato's ‘Cave Allegory’ from The Republic as a model for clinical research in PCOS. Are clinical researchers of PCOS fixated on the shadows of surrogate variables on the wall, or are they prepared to turn to the diseases on the rampart when seeking the truth of proposed treatments? From Saenredam after Cornelis Cornelisz, The Cave of Plato, Engraving, 1604 (London, British Museum).

 
Arguably the best-documented correlation between a surrogate outcome and a primary clinical end-point is that between ovulation and pregnancy in women with PCOS. There is good correlation between the increase in ovulation frequency with clomiphene citrate and the chance of pregnancy in women with PCOS, as large case series supports (Imani et al., 2002Go) and a Cochrane meta-analysis confirms (Hughes et al., 2000Go). Several reviews have also suggested a similar benefit with metformin (Barbieri, 2003Go; Haas et al., 2003Go; Harborne et al., 2003Go), and a recent Cochrane meta-analysis showed a significant treatment benefit of metformin in inducing ovulation in women with PCOS (Lord et al., 2003Go).

However, ovulation should not be equated with pregnancy, as a host of other factors may affect both pregnancy and the true outcome of interest: a healthy live born infant (Figure 3). For example, in the published meta-analysis of metformin in women with PCOS, the clinical pregnancy rate comparing metformin with placebo did not show evidence of benefit [odds ratio 2.76; 95% confidence interval (CI) 0.85–8.98; P=0.09; although it should be noted that in the trials comparing clomiphene with metformin compared with clomiphene alone, a significant pregnancy benefit was noted for clomiphene and metformin: odds ratio 4.40; 95% CI 1.96–9.85; Lord et al., 2003Go]. Factors both extrinsic (such as male factor, tubal factor or maternal age) and intrinsic to PCOS can affect the chance for pregnancy. These intrinsic factors include the quality of the PCOS oocyte (Ashkenazi et al., 1995Go; Chian et al., 1999Go), PCOS endometrial abnormalities that may affect implantation (Tuckerman et al., 2000Go) or other factors that may lead to an increased miscarriage risk in women with PCOS (Sagle et al., 1988Go; Rai et al., 2000Go).



View larger version (18K):
[in this window]
[in a new window]
 
Figure 3. The inverted pyramid of ovulation to liveborn healthy infant in women with PCOS seeking pregnancy. There are multiple areas where the pathophysiology of PCOS may adversely effect the take-home baby rate.

 
Additionally, the effect of pregnancy complications reputed to be more common in women with PCOS such as pre-eclampsia (Fridstrom et al., 1999Go) or gestational diabetes (Holte et al., 1998Go) may also affect the true outcome of interest, a viable, term infant. Obesity, common in women with PCOS, is an independent risk factor for stillbirth (Froen et al., 2001Go). Thus, ovulation is far removed on the casual pathway from a live birth, diminishing its value as a surrogate marker for pregnancy in women with PCOS. Concern about these potential adverse events, and ensuing litigation, has discouraged pharmaceutical companies from pursuing clinical trials for this secondary indication for insulin sensitizers in women with PCOS.

Another primary clinical outcome of interest, hirsutism, has been studied extensively in women with and without PCOS. It is interesting to note that in the largest clinical trial of women with PCOS (in which women were diagnosed on the basis of the NICHD recommended diagnostic criteria of unexplained hyperandrogenemic chronic anovulation; Zawadski and Dunaif, 1992Go), ~50% of the women did not have clinically recognized hirustism (Azziz et al., 2001Go). There is a lack of well-validated surrogate measures of hirsutism, and even uncertainty as to optimal methods for quantifying changes in hirsutism in response to treatment (Barth, 1996Go). Hirsutism scores are notoriously subjective (Holdaway et al., 1985Go), and even the most frequently utilized standard for hirsutism scores, the modified Ferriman–Gallwey score, relies partially on non-midline, non-androgen-dependent body hair to make the diagnosis (Hatch et al., 1981Go). In PCOS, the ability to discriminate unwanted excess hair with a diffuse distribution (hypertrichosis) from unwanted excess hair in androgen-dependent locations (hirsutism) is important, since reduction in excess androgen would be unlikely to improve hypertrichosis.

For example, other scales have focused on midline hair (Lorenzo, 1970Go; Derksen et al., 1993Go), and the best discrimination between a control population and a hirsute population has been found using the sum of the scores for four regions: upper lip, chin, lower abdomen and thighs (Derksen et al., 1993Go). Interobserver coefficient of variation between two observers was good (Derksen et al., 1993Go). Single site assessment of chin or lower abdomen has been found to be sensitive, but to have poor specificity in the larger population (positive predictive value in the general population of <60%) (Knochenhauer et al., 2000Go).

Despite these caveats about the difficulty of assessing hirsutism, there are acceptable methods for achieving an FDA indication for treating hirsutism. The approval of eflornithine hydrochloride cream for the treatment of female hirsutism by the FDA was based on a physician's global assessment scale, evaluating facial hair 48 h after shaving on treatment compared with placebo (Hickman et al., 2001Go). The methodology of this scale involved both objective and subjective measures of hirsutism (Hickman et al., 2001Go). This example provides inspiration for further investigation of medications that improve hirsutism in women with PCOS.


    Critique of other primary outcomes in women with PCOS
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
In the case where there is some evidence that surrogate markers may correlate well with the disease and response to treatment, such as C-reactive protein (Ridker et al., 2003Go), there is a lack of epidemiological evidence linking women with PCOS to an increased risk for the primary clinical outcome, i.e. cardiovascular events (Solomon, 1999Go). Models can be created to identify an increased risk profile in women with PCOS (Dahlgren et al., 1992Go), but this does not necessarily predict events in this subset of women. Thus, while it is clear that women with PCOS have multiple risk factors for cardiovascular disease, the evidence suggesting an increased number of events is still lacking (Legro, 2003Go).

There are also diseases where there are neither surrogate markers nor evidence of increased risk of the disease in women with PCOS. An example of this is endometrial cancer, and its precursor, atypical endometrial hyperplasia. There is little epidemiological data to suggest that women with PCOS per se are at increased risk for developing endometrial cancer (Hardiman et al., 2003Go). Again, the majority of data linking women with PCOS to endometrial cancer are case reports or case series (Jackson and Dockerty, 1957Go), or retrospective case–control studies identifying risk factors within the cases with endometrial cancer, such as chronic anovulation (Coulam et al., 1983Go) or hirsutism (Dahlgren et al., 1991Go) (i.e. chronic anovulation) as synonymous with PCOS. Thus, there is evidence that women with PCOS have risk factors for endometrial cancer (i.e. chronic anovulation, obesity and diabetes) without the firm epidemiological data supporting events.


    What are the feasible clinical end-points of interest in women with PCOS?
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
This critique of surrogate variables is not meant to discourage their use in clinical trials. They are of great importance in translational clinical research that examines novel hypotheses regarding the pathophysiology of PCOS, or in pilot trials of potential treatment agents. However, these do not always provide us with the evidence for managing these patients clinically. Large-scale multicentre trials in women with PCOS should focus on a primary clinical end-point and not a surrogate marker. Because there are scant data on the prevalence rates of endometrial cancer and cardiovascular disease in women with PCOS, and perhaps more importantly even fewer data about conversion rates over time in this at-risk population, it would be difficult to design a clinical trial around these end-points, without a generous estimate of baseline risk and conversion rates.

An intriguing area of further research would be a diabetes prevention trial in women with PCOS. There is a high prevalence of glucose intolerance (~40%) in these women (Ehrmann et al., 1999Go; Legro et al., 1999Go), present even in adolescence (Palmert et al., 2002Go), which is a strong modifiable risk factor for developing type 2 diabetes. Small studies in women with PCOS have shown high conversion rates to type 2 diabetes over time (Ehrmann et al., 1999Go; Norman et al., 2001Go). Further intervention trials, including both lifestyle and pharmaceutical interventions, have shown consistently a significant decline in the conversion to type 2 diabetes in women with impaired glucose tolerance (Chiasson et al., 1998Go; Buchanan et al., 2002Go; Knowler et al., 2002Go). This is an outcome that may prove feasible in a future trial in women with PCOS (although it should be noted that even the development of type 2 diabetes is a surrogate end-point—the true clinical end-points would be morbidity and mortality from complications of diabetes). As the above discussion suggests, the two clinical outcomes that appear most feasible as the end-point of a clinical trial are hirsutism and live birth.


    Conclusions
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
Implementing clinical trials in women with PCOS has been hampered by the lack of definitive diagnostic criteria, as well as by an emphasis on surrogate variables instead of primary clinical end-points. Most surrogate variables have not been validated as predictive of the clinical events of interest in these women. Although ovulation has frequently been studied as the end-point of clinical trial in women with PCOS, it is relatively far removed on the causal pathway from a live birth, and the two should not be viewed as the same end-point. Pregnancy and an improvement in hirsutism are clinical end-points that have been studied successfully in the past, and are being evaluated in ongoing clinical trials in women with PCOS. Many other clinical end-points, such as endometrial cancer and cardiovascular disease, owing to their rare occurrence in premenopausal women with PCOS, may not be suitable as the primary outcome variable of clinical studies; however, others, such as the prevention of diabetes, might be. Future multicentre trials in women with PCOS should focus on primary clinical end-points.


    Acknowledgements
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
This work was supported by PHS grants U10 HD 38992 and U01 HD 38997. This paper was reviewed and commented on by members of the Steering Committee of the Reproductive Medicine Network: Peter McGovern, MD, UMNDJ, Newark, NJ; William Schlaff, MD, University of Colorado School of Medicine, Denver, CO; Sandra Carson, Baylor College of Medicine, Bruce Carr, MD, UT Southwestern School of Medicine, Dallas, TX; Michael Steinkampf, MD, UAB School of Medicine, Birmingham, AL; Christos Coutifaris, MD, PhD, University of Pennsylvania, Philadelphia, PA; Linda Giudice, MD, Stanford University School of Medicine, Stanford, CA; Phyllis Leppert, MD, PhD, Reproductive Sciences Branch, NICHD, Rockville, MD, USA.


    References
 Top
 Abstract
 Introduction
 Potential diagnostic criteria in...
 Definition of surrogate end...
 Discussion of surrogate end...
 Critique of surrogate end-points
 Surrogate end-points in PCOS
 Critique of surrogate end-points...
 Critique of other primary...
 What are the feasible...
 Conclusions
 Acknowledgements
 References
 
Adams J, Polson DW and Franks S (1986) Prevalence of polycystic ovaries in women with anovulation and idiopathic hirsutism. BMJ 293, 355–359.[Medline]

American Diabetic Association (1998) Consensus Development Conference on Insulin Resistance. Diabetes Care 21, 310–314.[Medline]

Anonymous (1989) Preliminary report: effect of encainide and flecainide on mortality in a randomized trial of arrhythmia suppression after myocardial infarction. The cardiac arrhythmia suppression trial (cast) investigators. N Engl J Med 321, 406–412.[Abstract]

Anonymous (1992) Effect of the antiarrhythmic agent moricizine on survival after myocardial infarction. The Cardiac Arrhythmia Suppression Trial II investigators. N Engl J Med 327, 227–233.[Abstract]

Ashkenazi J, Farhi J, Orvieto R, Homburg R, Dekel A, Feldberg D and Ben Rafael Z (1995) Polycystic ovary syndrome patients as oocyte donors: the effect of ovarian stimulation protocol on the implantation rate of the recipient. Fertil Steril 64, 564–567.[Medline]

Asuncion M, Calvo RM, San Millan JL, Sancho J, Avila S and Escobar-Morreale HF (2000) A prospective study of the prevalence of the polycystic ovary syndrome in unselected Caucasian women from Spain. J Clin Endocrinol Metab 85, 2434–2438.[Abstract/Free Full Text]

Azziz R, Ehrmann D, Legro RS, Whitcomb RW, Hanley R, Fereshetian AG, O'Keefe M and Ghazzi MN and the PCOS/Troglitazone Study Group, (2001) Troglitazone improves ovulation and hirsutism in the polycystic ovary syndrome: a multicenter, double blind, placebo-controlled trial. Clin Endocrinol Metab 86, 1626–1632.[CrossRef]

Balen A and Michelmore K (2002) What is polycystic ovary syndrome? Are national views important? Hum Reprod 17, 2219–2227.[Abstract/Free Full Text]

Barbieri RL (2003) Metformin for the treatment of polycystic ovary syndrome. Obstet Gynecol 101, 785–793.[Abstract/Free Full Text]

Barth JH (1996) How robust is the methodology for trials of therapy in hirsute women? Clin Endocrinol 45, 379–380.[CrossRef][Medline]

Bloomgarden ZT (2003a) American Association of Clinical Endocrinologists (AACE) consensus conference on the insulin resistance syndrome: Washington, DC. Diabetes Care 26, 933–939.[Free Full Text]

Bloomgarden ZT (2003b) American Association of Clinical Endocrinologists (AACE) consensus conference on the insulin resistance syndrome: Washington, DC. Diabetes Care 26, 1297–1303.[Free Full Text]

Bo S, Menato G, Gallo ML, Bardelli C, Lezo A, Signorile A, Gambino R, Cassader M, Massobrio M and Pagano G (2004) Mild gestational hyperglycemia, the metabolic syndrome and adverse neonatal outcomes. Acta Obstet Gynecol Scand 83, 335–340.[CrossRef][Medline]

Boots LR, Potter S, Potter D and Azziz R (1998) Measurement of total serum testosterone levels using commercially available kits: high degree of between-kit variability. Fertil Steril 69, 286–292.[CrossRef][Medline]

Buchanan TA, Xiang AH, Peters RK, Kjos SL, Marroquin A, Goico J, Ochoa C, Tan S, Berkowitz K, Hodis HN et al. (2002) Preservation of pancreatic beta-cell function and prevention of type 2 diabetes by pharmacological treatment of insulin resistance in high-risk Hispanic women. Diabetes 51, 2796–2803.[Abstract/Free Full Text]

Chian RC, Gulekli B, Buckett WM and Tan SL (1999) Priming with human chorionic gonadotropin before retrieval of immature oocytes in women with infertility due to the polycystic ovary syndrome. N Engl J Med 341, 1624–1626.[Free Full Text]

Chiasson JL, Gomis R, Hanefeld M, Josse RG, Karasik A and Laakso M (1998) The STOP-NIDDM Trial: an international study on the efficacy of an alpha-glucosidase inhibitor to prevent type 2 diabetes in a population with impaired glucose tolerance: rationale, design, and preliminary screening data. Study to prevent non-insulin-dependent diabetes mellitus. Diabetes Care 21, 1720–1725.[Abstract]

Coulam CB, Annegers JF and Kranz JS (1983) Chronic anovulation syndrome and associated neoplasia. Obstet Gynecol 61, 403–407.[Abstract]

Cushman M, Legault C, Barrett-Connor E, Stefanick ML, Kessler C, Judd HL, Sakkinen PA and Tracy RP (1999) Effect of postmenopausal hormones on inflammation-sensitive proteins: the Postmenopausal Estrogen/Progestin Interventions (PEPI) study. Circulation 100, 717–722.[Abstract/Free Full Text]

Dahlgren E, Friberg LG, Johansson S, Lindstrom B, Oden A, Samsioe G and Janson PO (1991) Endometrial carcinoma; ovarian dysfunction—a risk factor in young women. Eur J Obstet Gynecol 41, 143–150.[Medline]

Dahlgren E, Janson PO, Johansson S, Lapidus L and Oden A (1992) Polycystic ovary syndrome and risk for myocardial infarction. Evaluated from a risk factor model based on a prospective population study of women. Acta Obstet Gynecol Scand 71, 599–604.[Medline]

DeFronzo RA, Tobin JD and Andres R (1979) Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Phys 237, E214–E223.

Derksen J, Moolenaar AJ, Van Seters AP and Kock DF (1993) Semiquantitative assessment of hirsutism in dutch women. Br J Dermatol 128, 259–263.[Medline]

Diamanti-Kandarakis E, Kouli CR, Bergiele AT, Filandra FA, Tsianateli TC, Spina GG, Zapanti ED and Bartzis MI (1999) A survey of the polycystic ovary syndrome in the Greek island of Lesbos: hormonal and metabolic profile. J Clin Endocrinol Metab 84, 4006–4011.[Abstract/Free Full Text]

Ehrmann DA, Barnes RB, Rosenfield RL, Cavaghan M and Imperial J (1999) Prevalence of impaired glucose tolerance and diabetes in women with polycystic ovary syndrome. Diabetes Care 22, 141–146.[Abstract]

Expert Panel on Detection Evaluation and Treatment of High Blood Cholesterol in Adults (2001) Executive summary of the third report of the National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III). JAMA 285, 2486–2497.[Free Full Text]

Flandre P and Saidi Y (1999) Estimating the proportion of treatment effect explained by a surrogate marker. Stat Med 18, 107–109.[CrossRef][Medline]

Fox R, Corrigan E, Thomas PA and Hull MG (1991) The diagnosis of polycystic ovaries in women with oligo-amenorrhoea: predictive power of endocrine tests. Clin Endocrinol 34, 127–131.[Medline]

Fridstrom M, Nisell H, Sjoblom P and Hillensjo T (1999) Are women with polycystic ovary syndrome at an increased risk of pregnancy-induced hypertension and/or preeclampsia? Hypertens Pregnancy 18, 73–80.[Medline]

Froen JF, Arnestad M, Frey K, Vege A, Saugstad OD and Stray-Pedersen B (2001) Risk factors for sudden intrauterine unexplained death: epidemiologic characteristics of singleton cases in Oslo, Norway, 1986–1995. Am J Obstet Gynecol 184, 694–702.[CrossRef][Medline]

Goldenthal, K (2001) Preventive HPV vaccine endpoints WebPage: http//www.fda.gov/ohrms/dockets/ac/01/slides/3805s1_01%20Goldenthal/tsld001.htm.

Gould AL, Rossouw JE, Santanello NC, Heyse JF and Furberg CD (1995) Cholesterol reduction yields clinical benefit. A new look at old data. Circulation 91, 2274–2282.[Abstract/Free Full Text]

Haas DA, Carr BR and Attia GR (2003) Effects of metformin on body mass index, menstrual cyclicity, and ovulation induction in women with polycystic ovary syndrome. Fertil Steril 79, 469–481.[CrossRef][Medline]

Harborne L, Fleming R, Lyall H, Norman J and Sattar N (2003) Descriptive review of the evidence for the use of metformin in polycystic ovary syndrome. Lancet 361, 1894–1901.[CrossRef][Medline]

Hardiman P, Pillay OS and Atiomo W (2003) Polycystic ovary syndrome and endometrial carcinoma. Lancet 361, 1810–1812.[CrossRef][Medline]

Hatch R, Rosenfield RL, Kim MH and Tredway D (1981) Hirsutism: implications, etiology, and management. Am J Obstet Gynecol 140, 815–830.[Medline]

Hickman JG, Huber F and Palmisano M (2001) Human dermal safety studies with eflornithine HCl 13.9% cream (Vaniqa), a novel treatment for excessive facial hair. Curr Med Res Opin 16, 235–244.[CrossRef][Medline]

Hine LK, Laird N, Hewitt P and Chalmers TC (1989) Meta-analytic evidence against prophylactic use of lidocaine in acute myocardial infarction. Arch Intern Med 149, 2694–2698.[Abstract]

Holdaway IM, Fraser A, Sheehan A, Croxson MS, France JT and Ibbertson HK (1985) Objective assessment of treatment response in hirsutism. Horm Res 22, 253–259.[Medline]

Holloway RG and Dick AW (2002) Clinical trial end points: on the road to nowhere? Neurology 58, 679–686.[Abstract/Free Full Text]

Holte J, Gennarelli G, Wide L, Lithell H and Berne C (1998) High prevalence of polycystic ovaries and associated clinical, endocrine, and metabolic features in women with previous gestational diabetes mellitus. J Clin Endocrinol Metab 83, 1143–1150.[Abstract/Free Full Text]

Hughes E, Collins J and Vandekerckhove P (2000) Clomiphene citrate for ovulation induction in women with oligo-amenorrhoea. Cochrane Database of Systematic Reviews, CD000056.

Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B and Vittinghoff E (1998) Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women. Heart and Estrogen/Progestin Replacement Study (HERS) research group. JAMA 280, 605–613.[Abstract/Free Full Text]

Imani B, Eijkemans MJ, de Jong FH, Payne NN, Bouchard P, Giudice LC and Fauser BC (2000) Free androgen index and leptin are the most prominent endocrine predictors of ovarian response during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. J Clin Endocrinol Metab 85, 676–682.[Abstract/Free Full Text]

Imani B, Eijkemans MJ, te Velde ER, Habbema JD and Fauser BC (2002) A nomogram to predict the probability of live birth after clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. Fertil Steril 77, 91–97.[CrossRef][Medline]

Jackson RL and Dockerty MB (1957) The Stein-Leventhal syndrome. Analysis of 43 cases with special reference to endometrial cancer. Am J Obstet Gynecol 73, 161–173.

Kahsar-Miller MD, Nixon C, Boots LR, Go RC and Azziz R (2001) revalence of polycystic ovary syndrome (PCOS) in first-degree relatives of patients with PCOS. Fertil Steril 75, 53–58.[CrossRef][Medline]

Kelly CC, Lyall H, Petrie JR, Gould GW, Connell JM and Sattar N (2001) Low grade chronic inflammation in women with polycystic ovarian syndrome. J Clin Endocrinol Metab 86, 2453–2455.[Abstract/Free Full Text]

Knochenhauer ES, Key TJ, Kahsar-Miller M, Waggoner W, Boots LR and Azziz R (1998) Prevalence of the polycystic ovary syndrome in unselected black and white women of the southeastern United States: a prospective study. J Clin Endocrinol Metab 83, 3078–3082.[Abstract/Free Full Text]

Knochenhauer ES, Hines G, Conway-Myers BA and Azziz R (2000) Examination of the chin or lower abdomen only for the prediction of hirsutism. Fertil Steril 74, 980–983.[CrossRef][Medline]

Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA and Nathan DM and the Diabetes Prevention Program Group, (2002) Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 346, 393–403.[Abstract/Free Full Text]

Koskinen P, Penttila TA, Anttila L, Erkkola R and Irjala K (1996) Optimal use of hormone determinations in the biochemical diagnosis of the polycystic ovary syndrome. Fertil Steril 65, 517–522.[Medline]

Legro RS (2003) Polycystic ovary syndrome and cardioivascular disease: A premature association? Endocr Rev 24, 302–312.[Abstract/Free Full Text]

Legro RS, Finegood D and Dunaif A (1998) A fasting glucose to insulin ratio is a useful measure of insulin sensitivity in women with polycystic ovary syndrome. J Clin Endocrinol Metab 83, 2694–2698.[Abstract/Free Full Text]

Legro RS, Kunselman AR, Dodson WC and Dunaif A (1999) Prevalence and predictors of risk for type 2 diabetes mellitus and impaired glucose tolerance in polycystic ovary syndrome: a prospective, controlled study in 254 affected women. J Clin Endocrinol Metab 84, 165–169.[Abstract/Free Full Text]

Legro RS, Kunselman AR and Dunaif A (2001) Prevalence and predictors of dyslipidemia in women with polycystic ovary syndrome. Am J Med 111, 607–613.[CrossRef][Medline]

Lin DY, Fleming TR and De Gruttola V (1997) Estimating the proportion of treatment effect explained by a surrogate marker. Stat Med 16, 1515–1527.[CrossRef][Medline]

Lord JM, Flight IH and Norman RJ (2003) Metformin in polycystic ovary syndrome: systematic review and meta-analysis. BMJ 327, 951–953.[Abstract/Free Full Text]

Lorenzo EM (1970) Familial study of hirsutism. J Clin Endocrinol Metab 31, 556–564.[Medline]

Manson JE, Hsia J, Johnson KC, Rossouw JE, Assaf AR, Lasser NL, Trevisan M, Black HR, Heckbert SR, Detrano R et al. (2003) Estrogen plus progestin and the risk of coronary heart disease. N Engl J Med 349, 523–534.[Abstract/Free Full Text]

Matsuda M and DeFronzo RA (1999) Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care 22, 1462–1470.[Abstract]

McArthur JW, Ingersall FM and Worcester J (1958) The urinary excretion of interstitial cell and folicle-stimulating hormone activity by women with diseases of the reproductive system. J Clin Endocrinol Metab 18, 1202–1205.

Michelmore KF, Balen AH, Dunger DB and Vessey MP (1999) Polycystic ovaries and associated clinical and biochemical features in young women. Clin Endocrinol 51, 779–786.[CrossRef][Medline]

the CONSORT Group (Consolidated Standards of Reporting Trials), Moher D, Schulz KF and Altman D (2001) The consort statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. JAMA 285, 1987–1991.[Abstract/Free Full Text]

Nestler JE and Jakubowicz DJ (1996) Decreases in ovarian cytochrome P450C17-alpha activity and serum free testosterone after reduction of insulin secretion in polycystic ovary syndrome. N Engl J Med 335, 617–623.[Abstract/Free Full Text]

Nestler JE, Jakubowicz DJ, Reamer P, Gunn RD and Allan G (1999) Ovulatory and metabolic effects of D-chiro-inositol in the polycystic ovary syndrome. N Engl J Med 340, 1314–1320.[Abstract/Free Full Text]

Norman RJ, Masters SC, Hague W, Beng C, Pannall P and Wang JX (1995) Metabolic approaches to the subclassification of polycystic ovary syndrome. Fertil Steril 63, 329–335.[Medline]

Norman RJ, Masters L, Milner CR, Wang JX and Davies MJ (2001) Relative risk of conversion from normoglycaemia to impaired glucose tolerance or non-insulin dependent diabetes mellitus in polycystic ovarian syndrome. Hum Reprod 16, 1995–1998.[Abstract/Free Full Text]

Palmert MR, Gordon CM, Kartashov A, Legro RS, Emans SJ and Dunaif A (2002) Screening for abnormal glucose tolerance in adolescents with polycystic ovary syndrome. J Clin Endocrinol Metab 87, 1017–1023.[Abstract/Free Full Text]

Prentice RL (1989) Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med 8, 431–440.[Medline]

Rai R, Backos M, Rushworth F and Regan L (2000) Polycystic ovaries and recurrent miscarriage—a reappraisal. Hum Reprod 15, 612–615.[Abstract/Free Full Text]

Raj SG, Thompson IE, Berger MJ, Talert LM and Taymor ML (1978) Diagnostic value of androgen measurements in polycystic ovary syndrome. Obstet Gynecol 52, 169–171.[Abstract]

Ridker PM, Hennekens CH, Buring JE and Rifai N (2003) C-reactive protein and other markers of inflammation in the prediction of cardiovascular disease in women. N Engl J Med 342, 836–843.[CrossRef]

Rodin DA, Bano G, Bland JM, Taylor K and Nussey SS (1998) Polycystic ovaries and associated metabolic abnormalities in Indian Subcontinent Asian Women. Clin Endocrinol 49, 91–99.[CrossRef][Medline]

Rosendaal FR, Helmerhorst FM and Vandenbroucke JP (2002) Female hormones and thrombosis. Arterioscler Thromb Vasc Biol 22, 201–210.[Abstract/Free Full Text]

Sagle M, Bishop K, Ridley N, Alexander FM, Michel M, Bonney RL, Beard RW and Franks S (1988) Recurrent early miscarriage and polycystic ovaries. BMJ 297, 1027–1028.[Medline]

Solomon CG (1999) The epidemiology of polycystic ovary syndrome. Prevalence and associated disease risks. Endocrinol Metab Clin 28, 247–263.

Stein IF and Leventhal ML (1935) Amenorrhea associated with polycystic ovaries. Am J Obstet Gynecol 29, 181–191.

Talbott E, Guzick D, Clerici A, Berga S, Detre K, Weimer K and Kuller L (1995) Coronary heart disease risk factors in women with polycystic ovary syndrome. Arterioscler Thromb Vasc Biol 15, 821–826.[Abstract/Free Full Text]

Talbott EO, Guzick DS, Sutton-Tyrrell K, McHugh-Pemu KP, Zborowski JV, Remsberg KE and Kuller LH (2000) Evidence for association between polycystic ovary syndrome and premature carotid atherosclerosis in middle-aged women. Arterioscler Thromb Vasc Biol 20, 2414–2421.[Abstract/Free Full Text]

Temple RJ (1995) A regulatory authority's opinion about surrogate endpoints. In Nimmo WS and Tucker G (eds) Clinical Measurement in Drug Evaluation. Wiley, New York, NY, USA.

The Rotterdam ESHRE/ASRM Sponsored PCOS Consensus Workshop Group (2004a) Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod 19, 41–47.[Abstract/Free Full Text]

The Rotterdam ESHRE/ASRM Sponsored PCOS Consensus Workshop Group (2004b) Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Fertil Steril 81, 19–25.[CrossRef][Medline]

Tremblay RR and Dube JY (1974) Plasma concentrations of free and non-TeBG bound testosterone in women on oral contraceptives. Contraception 10, 599–605.[Medline]

Tuckerman EM, Okon MA, Li T and Laird SM (2000) Do androgens have a direct effect on endometrial function? An in vitro study. Fertil Steril 74, 771–779.[CrossRef][Medline]

Urbanek M, Legro RS, Driscoll DA, Azziz R, Ehrmann DA, Norman RJ, Strauss JF, 3rd, Spielman RS and Dunaif A (1999) Thirty-seven candidate genes for polycystic ovary sydnrome: Strongest evidence for linkage is with follistatin. Proc Natl Acad Sci USA 96, 8573–8578.[Abstract/Free Full Text]

Velazquez EM, Mendoza SG, Wang P and Glueck CJ (1997) Metformin therapy is associated with a decrease in plasma plasminogen activator inhibitor-1, lipoprotein(a), and immunoreactive insulin levels in patients with the polycystic ovary syndrome. Metabolism 46, 454–457.[Medline]

Vermeulen A, Verdonck L and Kaufman JM (1999) A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 84, 3666–3672.[Abstract/Free Full Text]

Wijeyaratne CN, Balen AH, Barth JH and Belchetz PE (2002) Clinical manifestations and insulin resistance (IR) in polycystic ovary syndrome (PCOS) among South Asians and Caucasians: is there a difference? Clin Endocrinol 57, 343–350.[CrossRef][Medline]

Wild RA (2002) Long-term health consequences of pcos. Hum Reprod Update 8, 231–241.[Abstract/Free Full Text]

Writing Group for the Women's Health Initiative Investigators (2002) Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the women's health initiative randomized controlled trial. JAMA 288, 321–333.[Abstract/Free Full Text]

Yarali H, Yildirir A, Aybar F, Kabakci G, Bukulmez O, Akgul E and Oto A (2001) Diastolic dysfunction and increased serum homocysteine concentrations may contribute to increased cardiovascular risk in patients with polycystic ovary syndrome. Fertil Steril 76, 511–516.[CrossRef][Medline]

Zawadski JK and Dunaif A (1992) Diagnostic criteria for polycystic ovary syndrome; towards a rational approach. (ed. Dunaif A, Givens JR, Haseltine F and Merriam GR), Polycystic Ovary Syndrome. Blackwell Scientific, Boston, MA, USA, pp. 377–384.

Submitted on February 3, 2004; accepted on April 22, 2004.