Insulin-sensitizing agents as primary therapy for patients with polycystic ovarian syndrome

Sonya Kashyap1,2,3,4, George A. Wells1 and Zev Rosenwaks2

1 Department of Epidemiology and Community Medicine, University of Ottawa, Ottawa, Ontario, Canada, 2 Center for Reproductive Medicine, Weill Sanford Medical College of Cornell University, New York Presbyterian Hospital, New York, NY 10021, USA and3 Ottawa Health Research Institute (OHRI)

4 To whom correspondence should be addressed. Email: skashyap{at}ottawahospital.on.ca


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
BACKGROUND: This paper is a systematic review of metformin versus clomiphene citrate (CC) in women with polycystic ovary syndrome (PCOS). METHODS: Meta-analysis Of Observational Studies in Epidemiology (MOOSE) and QUality Of Reporting Of Meta-analyses (QUOROM) guidelines were followed. A systematic computerized literature search was done of seven bibliographic databases. Inclusion criteria included cohort and randomized controlled trials (RCT) of women with PCOS and the following medications: metformin versus placebo; metformin versus CC; metformin plus CC versus placebo plus CC. Rev-man 4.1 and Metaview 4.0 were used to analyse data. Relative risk (RR) estimates were presented. A {chi}2-test determined the significance of the association. Heterogeneity was determined by the Cochran Q-test. RESULTS: Metformin was 50% better than placebo for ovulation induction in infertile PCOS patients [RR 1.50; 95% confidence interval (CI) 1.13, 1.99]. Metformin was also of benefit in non-infertile (i.e. patients with PCOS who were not complaining of infertility) PCOS patients for cycle regulation compared to placebo (RR 1.45; CI 1.11, 1.90). Metformin was not of confirmed benefit versus placebo for achievement of pregnancy (RR 1.07; CI 0.20, 5.74). Metformin plus CC may be 3-4-fold superior to CC alone for ovulation induction (RR 3.04; CI 1.77, 5.24) and pregnancy (RR 3.65; CI 1.11, 11.99) in women with PCOS. CONCLUSIONS: Metformin is effective for ovulation induction and cycle regulation in this group of patients. Metformin plus CC appears to be very effective for achievement of pregnancy compared to CC alone. No RCTs directly compare metformin to CC but the need for such a trial exists.

Key words: clomiphene citrate/insulin sensitizers/meta-analysis/metformin/polycystic ovarian syndrome


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Polycystic ovary syndrome (PCOS) has traditionally been treated with ovulation-inducing medications such as clomiphene citrate (CC). More recently, research understanding the relationship between PCOS and insulin resistance has led to the use of insulin-sensitizing medications, such as metformin, for ovulation induction. However, despite widespread acceptance and zealous use of such medications, metformin has never been compared to CC in a direct, randomized controlled trial (RCT). The following systematic review will provide a comprehensive assessment of the available literature in preparation for such a study.

Theoretically, metformin reduces the insulin response by decreasing hepatic gluconeogenesis and reducing androgen levels, which allow resumption of normal menstrual cyclicity. Metformin should not confer the same risks of ovarian hyperstimulation and multiple pregnancy as CC since metformin returns patients to spontaneous ovulation not super ovulation. Also, metformin should not have the same negative effects on the cervical mucus and endometrium as CC. Metformin was first suggested as a treatment for ovulation induction in women with PCOS in the early 1990s. The first reports demonstrated resumption of menstrual cyclicity in 21 of 22 patients with PCOS (Velazquez et al., 1994Go). Since then, many studies have been conducted on small samples of women with PCOS, with the primary outcomes ranging from alteration of metabolic profiles to reinstitution of ovulation (Glueck et al., 1999Go). Most of these studies were observational and many retrospective. Several studies have demonstrated positive effects on the surrogate outcomes of metabolites such as androgen level and fasting insulin levels as well as ovulation (Moghetti et al., 2000Go). Also, none of the previous studies used pregnancy as a primary outcome.

The objective of this systematic review is to evaluate the available literature, RCT and cohort studies, regarding the use of CC versus metformin for induction of ovulation and achievement of pregnancy. A quantitative summary is provided only where appropriate. This study was done in preparation for an RCT investigating CC versus metformin which was started at our centre in August 2002. Subsequently, our study has been terminated but the Reproductive Medicine Network (funded by the NICHD) has been conducting a RCT investigating metformin, CC, and metformin plus CC. Thirteen centres are involved in the recruitment of 768 patients with oligomenorrhoea, elevated testosterone, normal semen parameters and regular intercourse (2–3 times/week) who desire pregnancy.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
One reviewer has completed all stages. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) and QUality Of Reporting Of Meta-analyses (QUOROM) guidelines were followed (Moher et al., 1999Go; Stroup et al., 2000Go).

Search strategy
We used an Ovid vendor and Polaris interface to conduct a computerized literature search of the following seven bibliographic databases: Medline, Premedline, Current Contents, Biological Abstracts, Cochrane Controlled Trials Register, Cochrane Database of Systematic Reviews, and EMBASE. The databases were searched for the last 25 years. MeSH headings and textwords, for databases without MeSH headings, were used. Adjacency operators and truncation were used. A preliminary search was conducted to maximize potential key words. We did not apply a study filter and the search was not limited by language or year of publication (Moher et al., 1996Go, 2000Go; Juni et al., 2002Go). The search was run every 3 weeks between August, 2002 and September, 2003 to identify new articles.

A hand search of the following journals was also completed: Fertility and Sterility, Human Reproduction, New England Journal of Medicine, Journal of Clinical Endocrinology, and The Lancet. Online indexing facilitated this process.

To avoid publication bias, we attempted to search for ‘grey literature’ (McAuley et al., 2000Go). Two content experts were contacted. No completed alternative studies were identified but an ongoing RCT was identified Results are not available (private communication, Reproductive Medicine Network, http://rmn.dcri.duke.edu/).

Ten years of conference proceedings for the American Society of Reproductive Medicine were also searched. The proceedings for the Canadian Fertility and Andrology Society were searched from 1998 onward. One abstract for a RCT was found (Singh et al., 2001Go).

The Cochrane Database was also searched for relevant papers. One protocol for a systematic review on the same topic was identified, but no completed review was found (Flight, 2002Go). The objective of the protocol was to assess multiple drugs of the same family and their effects on PCOS symptoms, sequelae (glucose intolerance, hypertension, cardiovascular disease), and adverse side-effects. This paper has, however, been published subsequently in both the Cochrane Database and the British Medical Journal and is considered in depth and compared to our review in the discussion (Lord et al., 2003Go).

Titles and abstracts were screened and articles retrieved if they passed the relevance filter or if there was uncertainty as to whether or not they were relevant. Bibliographies of review articles, systematic reviews and retrieved studies were also searched for candidate articles (Jadad et al., 1998Go). Retrieved articles were then reviewed for inclusion/exclusion criteria. Those articles that met the criteria were then kept for critical appraisal and data collection.

Reviewers were not blinded at any point to the authors or sources of publication as the evidence for such blinding is weak and the reviewers were likely to be previously familiar with some of the literature (Berlin, 1997Go).

Inclusion/exclusion criteria
Randomized controlled trials
Study population. The study population consists of women with primary or secondary infertility, between the ages of 18 and 40 years, who have been diagnosed with PCOS by the following characteristics: chronic oligo-ovulation (menstrual cycles less frequent than every 35 days or fewer than six periods per year); infertility; and one or more of the following characteristics: chronic hyperandrogenism (biochemical elevated testosterone, dehydroepiandrosterone, or androstenesdione levels or clinical hirustism or acne); increased LH/FSH ratio >2.5; or ultrasound criteria of PCOS. The study and control groups should have no other infertility diagnosis and preferably would have a documented normal semen analysis.

Intervention: We looked for the following comparisons: Metformin versus placebo; metformin versus CC; metformin plus CC versus placebo plus CC.

Outcome: Two outcomes are assessed: ovulation as determined by serum progesterone level, and pregnancy as determined by urinary or serum {beta}hCG.

Cohort studies
Similar inclusion/exclusion criteria were applied but no cohort studies were found that met the criteria.

For studies that resulted in multiple publications, only the most recent or most complete publication was used. Relevant data were collected onto a pre-formed, standard data extraction sheet. The following were recorded: study characteristics (source, language, year, and design); subject characteristics (definition and selection of controls, study subjects, available information on confounders); intervention/exposure information (drug doses and duration); and outcome assessment (methods of ascertainment of exposure or outcome, and time to assessment of outcome).

The Jadad scale was used for quality assessment of RCT (Jadad et al., 1996Go). Quality assessment involves evaluation of patient selection, assessment of exposure or outcomes, administration of interventions, and controls for confounding factors.

Quantitative data synthesis
Where appropriate, Rev-man 4.1 and Metaview 4.0 have been used to analyse data. A relative risk (RR) estimate, with confidence intervals (CI), was extracted from RCT. We reported the fixed effects model since results did not differ from a random effects model. A {chi}2-test was done to determine the significance of the association. Heterogeneity is determined by the Cochran Q-test. These results are illustrated graphically in the form of a Forrest plot. Funnel plots have been constructed to represent the likelihood of publication bias. The precision of the search is determined by the formula of Normand.

Where appropriate, sensitivity analysis and subgroup analysis were done to determine the significance of contributing factors to the overall results.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Overall precision of the search strategy was 0.76% (8/1057). Recall was 88% (7/8) (Figure 1).



View larger version (23K):
[in this window]
[in a new window]
 
Figure 1. A flow diagram showing the number of publications identified in the literature search. RCTS = randomized controlled trials. Generated from Meta-view 4.0.

 
Language of publication
Although we did not limit the language of publication, all retrieved articles were in English (Moher et al., 2000Go; Juni et al., 2002Go).

Year of publication
All included RCT were published between 1998 and 2002 (Nestler et al., 1998Go; Moghetti et al., 2000Go; Ng et al., 2001Go; Singh et al., 2001Go; Sturrock et al., 2001Go; Vandermolen et al., 2001Go; Fleming et al., 2002Go; Kocak et al., 2002Go).

Quality of publication
The Jadad quality scale for RCTs was applied to all included studies and is shown in Table I (Clark et al., 1999Go).


View this table:
[in this window]
[in a new window]
 
Table I. Quality assessment by Jadad scale

 
No RCTs were excluded for multiple publication. Studies were assessed for comparability of study population by age, body mass index, fertility, and method of diagnosis of PCOS. We also reviewed investigation for potential confounders, compliance, and contamination or cross-over. None of the studies commented on contamination or cross-over; only one study mentioned compliance but that study did not describe the methods used to assess compliance (Vandermolen et al., 2001Go). One study described drop-outs but did not discuss compliance methods or if compliance was measured in continuing subjects (Fleming et al., 2002Go).

All studies, except Singh et al. (2001)Go, investigated metformin versus placebo. All studies except three used pregnancy as an outcome but only one abstract assessed pregnancy as a primary outcome (Nestler et al., 1998Go; Moghetti et al., 2000Go; Ng et al., 2001Go). All studies except three had infertile, PCOS women as their study population (Nestler et al., 1998Go; Moghetti et al., 2000Go; Fleming et al., 2002Go). Several studies tried to compare metformin plus CC to CC alone or with placebo (Nestler et al., 1998Go; Ng et al., 2001Go; Vandermolen et al., 2001Go; Kocak et al., 2002Go). However, all of these studies did so in a sequential manner in those patients who failed to conceive or ovulate with metformin. The duration of intervention and follow-up varies from 1 to 6 months. Table IITable II illustrates the characteristics of these studies. Table III outlines objectives for each study.


View this table:
[in this window]
[in a new window]
 
Table II. Characteristics of randomized controlled trials

 

View this table:
[in this window]
[in a new window]
 
Table II. Characteristics of randomized controlled trials (continued)

 

View this table:
[in this window]
[in a new window]
 
Table III. Study objectives of randomized controlled trials

 
Metformin induced ovulation in 47% of patients. Figures 2 and 3 indicate that metformin is 50% better than placebo for ovulation induction in infertile PCOS patients (RR 1.50) but this benefit is not necessarily improved with longer duration of therapy (>3 months, RR 1.37). Metformin is also of benefit in non-infertile PCOS patients (i.e. patients with PCOS who were not complaining of infertility) for cycle regulation compared to placebo (Figure 3; RR 1.45 CI 1.11, 1.90). To date, metformin is not of confirmed benefit versus placebo for achievement of pregnancy (Figure 2; RR 1.07 CI 0.20, 5.74). This may be due to the fact that most studies used ovulation as their primary outcome, those studies that assessed pregnancy had a short follow-up time to pregnancy, and most studies had small sample sizes.



View larger version (29K):
[in this window]
[in a new window]
 
Figure 2. Comparison of metformin versus placebo in infertile polycystic ovarian syndrome (PCOS) patients. The upper panel shows the outcome ovulation and the lower panel the outcome pregnancy. RR = relative risk; CI = confidence interval. Generated from Meta-view 4.0.

 


View larger version (25K):
[in this window]
[in a new window]
 
Figure 3. Comparison of metformin versus placebo with outcome ovulation in randomized controlled trials >3 months treatment duration. The upper panel shows infertile polycystic ovarian syndrome (PCOS) patients and the lower panel not necessarily (nec) infertile patients with PCOS. RR = relative risk; CI = confidence interval. Generated from Meta-view 4.0.

 
The analysis of metformin plus CC versus CC alone or with placebo is not properly represented here since the study design for this component is not a true RCT or cross-over design but rather a sequential study design. The nature of infertility therapy makes cross-over trials an inappropriate design. In order to conduct a cross-over trial, the two therapies being compared must be able to be ‘washed out’ within a short window of time between arms, the outcome must also be completely reversible within this window so that the patient returns to baseline, and the chance of successful outcome should not alter with time such that one therapy has a disadvantage over the other by being administered in the later arm or that the chance of success varies with time. Nevertheless, we did do a quantitative analysis on the data supplied to provide some assessment of the above interventions for ovulation and pregnancy (Figure 4). Given the stated limitations of such a study design, we found that metformin plus CC may be superior to CC alone or with placebo with regards to ovulation (RR 3.04, CI 1.77, 5.24) and pregnancy (RR 3.65, CI 1.11, 11.99).



View larger version (27K):
[in this window]
[in a new window]
 
Figure 4. Comparison of metformin + clomiphene citrate (CC) versus placebo + CC in infertile patients. The upper panel shows the outcome ovulation and the lower panel the outcome pregnancy. RR = relative risk. Generated from Meta-view 4.0.

 
Funnel plots for each comparison did not suggest obvious publication bias but only two to five points were available for each analysis.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Specific criteria for the diagnosis of PCOS remain elusive. The clinical manifestations of hyperandrogenism may not always accompany elevated androgen levels since genetic susceptibility to these hormones at the target organ is also required. Hyperinsulinaemia has been documented to occur in 50–100% of obese PCOS patients and as high as 22% of lean PCOS patients (Dale et al., 1992Go). Extreme cases of hyperinsulinaemia may present with hyperandrogenism, insulin resistance, and acanthosis nigricans (HAIR-AN syndrome). The most recent NIH consensus conference on PCOS in 1990 defined the syndrome as chronic anovulation with hyperandrogenism. LH:FSH ratios, ultrasound criteria and hyperinsulinaemia were hitherto not included in this definition. The recent 2003 Rotterdam ESHRE/ASRM consensus on PCOS attempted to delineate a clearer definiton of PCOS but also stated that ‘PCOS remains a syndrome and no single diagnostic criterion (such as hyperandrogenism or PCO) is sufficient for clinical diagnosis. PCOS also remains a diagnosis of exclusion.’ Nevertheless, they stated that the revised diagnostic criteria include two of the following three findings: (i) oligo- or anovulation, (ii) clinical and/or biochemical signs of hyperandrogenism, (iii) polycystic ovaries and exclusion of other aetiologies (ESHRE and ASRM, 2004Go). Clearly, as the Rotterdam proceedings suggested, the definition of PCOS may be flexible but the criteria employed in a study should directly address the interventions and outcomes assessed. The revised Rotterdam criteria are appropriate here. Ovulation and pregnancy rates, not metabolic parameters, were assessed in this study.

The evidence that hyperinsulinaemia causes hyperandrogenism, which in turn affects fertility, is more convincing than arguments that hyperandrogenism causes hyperinsulinaemia. Following insulin infusion, adrenal androgen levels rise (Elkind-Hirsch et al., 1991Go). Conversely, treatment of hyperinsulinaemia with insulin-sensitizing agents or weight loss [5% or body mass index (BMI) <27 kg/m2] decreases androgen levels (Loverro et al., 2002Go; Mitkov et al., 2002Go). Medically or surgically induced menopause decreases circulating androgen levels without any effect on serum insulin levels (Nagamani et al., 1986Go).

Previously, primary therapy PCOS patients involved ovarian drilling, possibly followed by ovulation induction with CC or gonadotrophins (Farquhar et al., 2001Go). Ovarian drilling was hypothesized to decrease ovarian androgen production. Currently, standard treatment for ovulation induction in these patients is CC. The success rate of CC varies; 80% ovulation and 30–40% pregnancy rates have been reported (Gorlitsky et al., 1978Go; Lunenfeld et al., 1991Go; Kousta et al., 1997Go; Imani et al., 1998Go, 1999Go). Seventy-five per cent of pregnancies are achieved within the first 3 months of an ovulatory dose of CC (Imani et al., 1998Go). Some authors have suggested that ovulation induction with CC for a period >6 months in properly selected patients with PCOS can provide a cumulative pregnancy rate >90% (Messinis and Milingos, 1997Go). The proposed mechanism of action is that CC is an anti-estrogen which leads to increased production of pituitary production of gonadotrophins (FSH and LH). CC does not address the hyperandrogenic or hyperinsulinaemic environment. CC also has an anti-estrogenic effect on the endometrial lining and cervical mucus. The incidence of multiple pregnancy (mostly twins) with CC is 4–10% (Kousta et al., 1997Go; Eijkemans et al., 2003Go). Although the 1% risk of ovarian hyperstimulation sydrome (OHSS) after gonadotrophin therapy may be much less with CC, it may still occur.

Metformin is a biguanide insulin sensitizer. It is labelled as class B in pregnancy. Sufficient human data are not available but the drug has not been associated with congenital defects in animals. It has been used in the treatment of women with diabetes mellitus II without negative effects on the fetus. Recently, small series of women who have taken metformin throughout pregnancy to prevent miscarriage and gestational diabetes have been published and there were no reported defects (Glueck et al., 2002Go; Jakubowicz et al., 2002Go). The mechanism of action is mainly the inhibition of hepatic gluconeogenesis. Metformin also increases peripheral glucose utilization and insulin sensitivity, but it is not associated with hypoglycaemia.

Recently, Lord et al. (2003)Go published meta-analyses with similar results in both the Cochrane Collaboration and the British Medical Journal. The protocol for their paper was identified during our search but the papers were published after the last search date. While their results concerning ovulation and pregnancy were similar to those reported here, six studies were included in both papers, our paper included two different studies and their paper included five different studies (Nestler and Jakubowicz, 1996Go; Moghetti et al., 2000Go; El-Biely and Habba, 2001Go; Jakubowicz et al., 2001Go; Sturrock et al., 2001Go; Malkawi and Qulban, 2002Go; Yarali et al., 2002Go). Two of these five papers were not available to us (El-Biely and Habba, 2001Go; Malkawi and Qulban, 2002Go). Another evaluated co-administration of metformin during recombinant (r) FSH treatment of patients with CC-resistant PCOS (Yarali et al., 2002Go). Two other papers had been initially identified but nothing in their title, abstract or methods suggested relevance (Nestler and Jakubowicz, 1996Go; Jakubowicz et al., 2001Go). Nevertheless, our results are quite similar (Table IV).


View this table:
[in this window]
[in a new window]
 
Table IV. Metformin in polycystic ovarian syndrome (PCOS): comparison of meta-analyses results

 
Meta-analyses are limited by biases introduced through individual studies as well as biases introduced through the processes of systematic review and quantitative summary. One consistent problem is that none of the studies were powered to assess the most important outcome, i.e. pregnancy. The follow-up varied from 1 to 6 months. While oligomenorrhoea/ovulation was generally a criterion, studies varied in their assessment and reporting of hyperandrogenaemia and CC resistance and this may have produced selection bias. The ‘cross-over design’ used in several studies to assess metformin plus CC versus CC was inappropriate. RCT also differed significantly in their reporting of methods used to randomize and in analysis methods (intention to treat versus per protocol). Contamination and cross-over were not reported. Not all studies evaluated infertile PCOS patients and when they did they did not always assess duration of infertility or other contributing factors (misclassification bias). We attempted to reduce publication, multiple publication, reporting, conformity and retrieval biases by comprehensive literature searches and searches for grey literature. Only one reviewer completed all steps so that extractor and recording bias was possible. We attempted to limit bias by extracting information on three separate occasions. RCT data are less prone to bias than observational data and by nature of the study design, differences in control versus study groups should balance once the study population has been defined.

No RCT directly compared metformin to CC for induction of ovulation and/or attainment of pregnancy. Cohort studies were insufficiently homogeneous to allow quantifiable summary. This review provides level 1a evidence regarding metformin versus placebo for ovulation induction and pregnancy. This study suggests that metformin is superior to placebo for ovulation induction in patients with PCOS but that this benefit is not more pronounced with longer therapy (i.e. >3 months). The definition of PCOS is very important in determining the group of patients who will have an optimal response. Ng et al. (2001)Go did not find a benefit with metformin. The patients in their study were not overweight (BMI <23 kg/m2) and also were not hyperandrogenaemic as opposed to the other studies where a benefit was found. In a well-defined group of PCOS patients who do not complain of infertility, metformin also has significant advantage over placebo for resumption of ovulation and regulation of menstrual cycles.

Nevertheless, the data so far do not demonstrate a benefit of metformin versus placebo when the outcome considered is pregnancy. The follow-up time to pregnancy was short and, in the quantitatively summarized studies, pregnancy was not the primary outcome nor were these studies powered to assess pregnancy as an outcome. When we compared metformin plus CC to CC or metformin alone, however, there appeared to be a significant benefit of the combination treatment for both ovulation and pregnancy in patients with PCOS who were both hyperandrogenemic and overweight. However, the comparison of metformin with CC versus placebo with CC was done in each of these studies as a ‘sequential study’ rather than a true RCT or cross-over study.

A need still exists to compare directly metformin and CC as first-line agents for ovulation induction and achievement of singleton pregnancy in patients with well-defined PCOS. PCOS patients may gain more than ovulation induction from metformin therapy, including reduced miscarriage rates, lower incidences of multiple pregnancies, gestational diabetes, and ovarian hyperstimulation, and longer term cardiovascular health benefits. The currently ongoing study (Pregnancy in Polycystic Ovarian Syndrome: PPCOS) is designed to answer the first question and has the potential to provide valuable insight into the clinical management of this elusive syndrome.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Adams J, Polson JW, Franks S (1986) Prevalence of polycystic ovaries in women with anovulation and idiopathic hirsutism. Br Med J 293, 355–359.[ISI][Medline]

Berlin JA (1997) Does blinding of readers affect the results of meta-analysis? Lancet 350, 185–186.[ISI][Medline]

Clark HD, Wells GA, Huet C, McAlister FA, Salmi LR, Fergusson D and Laupacis A (1999) Assessing the reliability of randomized trials: reliability of the Jadad scale. Controlled Clin Trials 20, 448–452.[CrossRef][ISI][Medline]

Dale P, Tanbor T and Vaaler S (1992) Body weight, hypersinulinemia, and gonadotropin levels in the polycystic ovarian syndrome: evidence of two distinct populations. Fertil Steril 58, 487–491.[ISI][Medline]

Eijkemans MJC, Imani B, Mulders AGMGJ, Habbema JDF and Fauser BCJM (2003) High singleton live birth rate following classical ovulation induction in normogonadotrophic anovulatory infertility (WHO 2). Hum Reprod 18, 2357–2362.[Abstract/Free Full Text]

El-Biely MM and Habba M (2001) The use of metformin to augment the induction of ovulation in obese infertile patients with polycystic ovary syndome. Mid East Fertil Soc J 6, 43–49.

Elkind-Hirsch K, Valdes C and McConnel T (1991) Androgen responses to acutely increased endogenous insulin levels in hyperandrogenic and normal cycling women. Fertil Steril 55, 486.[ISI][Medline]

ESHRE and ASRM (2004) Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome*1. Fertil Steril 81, 19–25.[CrossRef][ISI][Medline]

Farquhar, C., Vandekerckhove, P., Lilford, R (2001) Laparoscopic "drilling" by diathermy or laser for ovulation induction in anovulatory polycystic ovary syndrome [update of Cochrane Database Syst Rev. 2000;(2): CD001122; PMID: 10796746]. Cochrane Database of Systematic Reviews CD001122.

Fleming R, Hopkinson ZE, Wallace M, Greer IA and Sattar N (2002) Ovarian function and metabolic factors in women with oligomenorrhea treated with metformin in a randomized double blind placebo-controlled trial. J Clin Endocrinol Metab 87, 569–574.[Abstract/Free Full Text]

Flight, N.R (2002) Insulin-sensitizing drugs for polycystic ovary syndrome (protocol). The Cochrane Library.

Glueck C, Wang P, Fontaine R, Tracy T and Sieve-Smith L (1999) Metformin-induced resumption of normal menses in 39 of 43 (91%) previously amenorrheic women with polycystic ovarian syndrome. Metab Clin Exp 48, 511–519.[ISI][Medline]

Glueck CJ, Wang P, Kobayashi S, Phillips H and Sieve-Smith L (2002) Metformin therapy throughout pregnancy reduces the development of gestational diabetes in women with polycystic ovary syndrome. Fertil Steril 77, 520–525.[CrossRef][ISI][Medline]

Gorlitsky G, Kase N and Speroff L (1978) Ovulation and pregnancy rates with clomiphene citrate. Obstet Gynecol 51, 265–269.[Abstract]

Imani B, Eijkemans MJC, te Velde ER, Habbema JDF and Fauser BCJM (1998) Predictors of patients remaining anovulatory during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. J Clin Endocrinol Metab 83, 2361–2365.[Abstract/Free Full Text]

Imani B, Eijkemans MJC, te Velde ER, Habbema JDF and Fauser BCJM (1999) Predictors of chances to conceive in ovulatory patients during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. J Clin Endocrinol Metab 84, 1617–1622.[Abstract/Free Full Text]

Jadad AR, Moore A, Carroll D, Jenkinson C, Reynolds JM, Gavaghan D and McQuay HJ (1996) Assessing the quality of reports of randomized clinical trials: is blinding necessary? Controlled Clin Trials 17, 1–12.[ISI][Medline]

Jadad AR, Moher D and Klassen TP (1998) Guides for reading and interpreting systematic reviews: II. How did the authors find the studies and assess their quality? Arch Pediatr Adolesc Med 152, 812–817.[Free Full Text]

Jakubowicz DJ, Seppala M, Jakubowicz S, Rodriguez-Armas O, Rivas-Santiago A, Koistinen H, Koistinen R and Nestler JE (2001) Insulin reduction with metformin increases luteal phase serum glycodelin and insulin-like growth factor-binding protein 1 concentrations and enhances uterine vascularity and blood flow in the polycystic ovary syndrome. J Clin Endocrinol Metab 86, 1126–1133.[Abstract/Free Full Text]

Jakubowicz DJ, Iuorno MJ, Jakubowicz S, Roberts KA and Nestler JE (2002) Effects of metformin on early pregnancy loss in the polycystic ovary syndrome. J Clin Endocrinol Metab 87, 524–529.[Abstract/Free Full Text]

Juni P, Holenstein F, Sterne J, Bartlett C and Egger M (2002) Direction and impact of language bias in meta-analyses of controlled trials: empirical study [comment]. Int J Epidemiol 31, 115–123.[Abstract/Free Full Text]

Kocak M, Caliskan E, Simsir C and Haberal A (2002) Metformin improves ovulatory rates, cervical scores, and pregnancy rates in clomiphene citrate-resistant women with polycystic ovary syndrome. Fertil Steril 77, 101–106.[CrossRef][ISI]

Kousta E, White DM and Franks S (1997) Modern use of clomiphene citrate in induction of ovulation. Hum Reprod Update 3, 359.[Abstract/Free Full Text]

Lord JM, Flight IHK and Norman RJ (2003) Metformin in polycystic ovary syndrome: systematic review and meta-analysis. Br Med J 327, 951–950.[Abstract/Free Full Text]

Loverro G, Lorusso F, De PagolaNicolardi G, Mei L and Selvaggi L (2002) Clinical and endocrinological effects of 6 months of metformin treatment in young hyperinsulinemic patients affected by polycystic ovarian syndrome. Gynecol Endocrinol 16, 217–224.[ISI][Medline]

Lunenfeld B, Pariente C, Dor J, Menashe Y, Seppala M, Mortman H and Insler V (1991) Modern aspects of ovulation induction. Ann NY Acad Sci 626, 207–216.[Medline]

Malkawi HY and Qulban HS (2002) The effect of metformin plus clomiphene citrate on ovulation and pregnancy rates in clomiphene-resistant women with polycystic ovary syndrome. Saudi Med J 23, 663–666.[ISI][Medline]

McAuley L, Pham B, Tugwell P and Moher D (2000) Does the inclusion of grey literature influence estimates of intervention effectiveness reported in meta-analyses? Lancet 356, 1228–1231.[CrossRef][ISI][Medline]

Messinis IE and Milingos SD (1997) Current and future status of ovulation induction in polycystic ovary syndrome. Hum Reprod Update 3, 235–253.[Abstract/Free Full Text]

Mitkov M, Simenov S and Pehlivanov B (2002) Effect of metformin treatment of polycystic ovary syndrome. Endocrinologia 7, 25–29.

Moghetti P, Castello R, Negri C, Tosi F, Perrone F, Caputo M, Zanolin E and Muggeo M (2000) Metformin effects on clinical features, endocrine and metabolic profiles, and insulin sensitivity in polycystic ovary syndrome: a randomized, double-blind, placebo-controlled 6-month trial, followed by open, long-term clinical evaluation. J Clin Endocrinol Metab 85, 139–145.[Abstract/Free Full Text]

Moher D, Fortin P, Jadad AR, Juni P, Klassen T, Le Lorier J, Liberati A, Linde K and Penna A (1996) Completeness of reporting of trials published in languages other than English: implications for conduct and reporting of systematic reviews. [comment]. Lancet 347, 363–366.[ISI][Medline]

Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D and Stroup DF (1999) Improving the quality of reports of meta-analyses of randomised controlled trials: the QUOROM statement. Quality of reporting of meta-analyses [comment]. Lancet 354, 1896–1900.[CrossRef][ISI][Medline]

Moher D, Pham B, Klassen TP, Schulz KF, Berlin JA, Jadad AR and Liberati A (2000) What contributions do languages other than English make on the results of meta-analyses? J Clin Epidemiol 53, 964–972.[CrossRef][ISI][Medline]

Nagamani M, Dinh T and Kelver M (1986) Hyperinsulinemia in hyperthecosis of the ovaries. Am J Obstet Gynecol 154, 384.[ISI][Medline]

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, Evans WS and Pasquali R (1998) Effects of metformin on spontaneous and clomiphene induced ovulation in the polycystic ovary syndrome. N Engl J Med 338, 1876–1880.[Abstract/Free Full Text]

Ng EHY, Wat NMS and Ho PC (2001) Effects of metformin on ovulation rate, hormonal and metabolic profiles in women with clomiphene-resistant polycystic ovaries:a randomized, double-blind placebo-controlled trial. Hum Reprod 16, 1625–1631.[Abstract/Free Full Text]

Singh I, Bedaiwy MA, Hatwal A, Kumar A and Agarwal A (2001) Increased pregnancy rates with metformin and clomiphene citrate in non-obese patients with polycystic ovary syndrome: prospective randomized study. Fertil Steril 76, S94.

Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D, Becker BJ, Sipe TA and Thacker SB (2000) Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. J Am Med Assoc 283, 2008–2012.[Abstract/Free Full Text]

Sturrock NDC, Lannon B and Fay TN (2001) Metformin does not enhance ovulation induction in clomiphene resistant polycystic ovary syndrome in clinical practice. Br J Pharmacol 53, 469–473.[CrossRef]

Vandermolen DT, Ratts VS, Evans WS, Stovall DW, Kauma SW and Nestler JE (2001) Metformin increases the ovulatory rate and pregnancy rate from clomiphene citrate in patients with polycystic ovary syndrome who are resistant to clomiphene citrate alone. Fertil Steril 75, 310–315.[CrossRef][ISI][Medline]

Velazquez E, Mendoza S, Hamer T, Sosa F and Glueck C (1994) Metformin therapy in polycystic ovarian syndrome reduces hyperinsulinema, insulin resistance, hyperandrogenemia, and systolic blood pressure, while facilitating normal menses and pregnancy. Metab Clin Exp 43, 647–654.[ISI][Medline]

Yarali H, Yildiz BO, Demirol A, Zeyneloglu HB, Yigit N, Bukulmez O and Koray Z (2002) Co-administration of metformin during rFSH treatment in patients with clomiphene citrate-resistant polycystic ovarian syndrome: a prospective randomized trial. Hum Reprod 17, 289–294.[Abstract/Free Full Text]

Submitted on February 9, 2004; resubmitted on May 4, 2004; accepted on July 6, 2004.