Hum. Reprod. Advance Access originally published online on January 29, 2007
Human Reproduction 2007 22(5):1200-1209; doi:10.1093/humrep/dem005
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Metformin versus oral contraceptive pill in polycystic ovary syndrome: a Cochrane review
1 Division of Obstetrics and Gynaecology, School of Women's and Children's Health, University of New South Wales, Royal Hospital for Women, Sydney, NSW, 2031, Australia 2 Department of Reproductive Medicine, Royal Hospital for Women, Sydney, NSW, Australia 3 IVFAustralia, Sydney, NSW, Australia 4 National Women's Department of Obstetrics and Gynaecology, University of Auckland, Auckland, New Zealand 5 Fertility Centre Scandinavia, Stockholm Storangsvagen 10, Stockholm, Sweden
6 To whom correspondence should be addressed at: Division of Obstetrics and Gynaecology, School of Women's and Children's Health, Level 1 Women's Health Institute, Royal Hospital for Women, Locked Bag 2000, Randwick, Sydney, NSW 2031, Australia. Fax: 61 2 9382 6444; E-mail: mfcostello{at}unsw.edu.au
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
BACKGROUND: The object of this review was to compare metformin versus oral contraceptive pill (OCP) treatment in polycystic ovary syndrome.
METHODS: A systematic review and meta-analysis employing the principles of the Cochrane Menstrual Disorders and Subfertility Group was undertaken.
RESULTS: Four randomized controlled trials (RCTs) (104 subjects) were included. Limited data demonstrated no evidence of a difference in effect between metformin and the OCP on hirsutism, acne or development of type 2 diabetes mellitus. There were no trials assessing diagnosis of cardiovascular disease or endometrial cancer. Metformin, in comparison with the OCP, was less effective in improving menstrual pattern [Peto odds ratio (OR) 0.08, 95% confidence interval (CI) 0.01–0.45) and in reducing the serum total testosterone level weighted mean difference (WMD) 0.54, 95% CI 0.22–0.86] but more effective in reducing fasting insulin (WMD –3.46, 95% CI – 5.39 to –1.52) and not increasing fasting triglyceride (WMD –0.48, 95% CI – 0.86 to –0.09) levels. Limited data demonstrated no evidence of a difference in effect between the two therapies on reducing fasting glucose or total cholesterol levels and severe adverse events.
CONCLUSIONS: The limited RCT evidence to date does not show adverse metabolic risk with the use of the OCP compared with metformin. Further long-term RCTs are required.
Key words: meta-analysis/metformin/oral contraceptive pill/polycystic ovary syndrome
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
Polycystic ovary syndrome (PCOS) is characterized by chronic anovulation and hyperandrogenism and affects
5–10% of women of reproductive age (Knochenhauer et al., 1998
; Diamanti-Kandarakis et al., 1999). PCOS is probably the most prevalent endocrinopathy in women (Homburg, 1996). PCOS is a heterogenous condition (clinically and biochemically) in affected women presenting in clinical practice seeking treatment for reproductive disorders such as menstrual cycle disturbances (oligomenorrhoea, amenorrhoea), infertility, hirsutism or acne (Balen and Michelmore, 2002). In addition, women with PCOS are thought to be at increased risk for endometrial cancer through chronic anovulation with consequent unopposed estrogen exposure (i.e. unopposed by progesterone) of the endometrium (Hardiman et al., 2003). However, it has recently become clear that PCOS is also linked to a number of metabolic disturbances, including type 2 (non-insulin-dependent) diabetes mellitus (T2DM) and possibly cardiovascular disease (CVD) (Ovalle and Azziz, 2002; Wild 2002).
The primary aetiology of PCOS is unknown (Balen, 2004). However, insulin resistance with compensatory hyperinsulinaemia is a prominent feature of the syndrome and appears to have a pathophysiological role in the hyperandrogenism of the disorder for both lean and obese women with PCOS (Dunaif et al., 1989). Hyperinsulinaemia stimulates both ovarian and adrenal androgen secretion directly and suppresses sex hormone-binding globulin synthesis from the liver, resulting in an increase in free, biologically active androgens. This excess in local ovarian androgen production augmented by hyperinsulinaemia causes premature follicular atresia and anovulation along with the other clinical manifestations of hyperandrogenism such as hirsutism and acne (Utiger, 1996).
Oral contraceptive pills (OCPs) have been the traditional medical therapy for the long-term treatment of PCOS in order to provide endometrial protection, regularize menses and to improve hirsutism and/or acne by reducing ovarian androgen production. However, recent limited and contradictory observational evidence has raised the concern that OCPs may reduce insulin sensitivity and glucose tolerance in PCOS women but there is no evidence that OCPs modify the risk of T2DM or CVD either negatively or positively (Diamanti-Kandarakis et al., 2003; Vrbikova and Cibula, 2005).
More recently, insulin-sensitizing drugs have been advocated as another long-term medical treatment option. Given the importance of hyperinsulinaemia in the development of hyperandrogenism and anovulation, it seems possible that insulin-sensitizing drugs may be useful in the restoration of normal endocrinological and clinical parameters of PCOS by lowering insulin secretion (Harborne et al., 2003a). Of the insulin-sensitizing drugs, metformin has been the one studied most widely and has the most reassuring safety profile (Nestler et al., 2002), particularly since the withdrawal of troglitazone from the US markets in 1999 by the Food and Drug Administration (FDA) of USA owing to numerous reports of fatal liver toxicity (Baillargeon et al., 2003; Ehrmann and Rychlik, 2003). Metformin enhances insulin sensitivity in both the liver, where it inhibits hepatic glucose production, and the peripheral tissue, where it increases glucose uptake and utilization into muscle tissue. By increasing insulin sensitivity, metformin reduces insulin resistance, insulin secretion and hyperinsulinaemia (Dunn and Peters, 1995).
Two recent systematic reviews have shown metformin to be superior to placebo in regularizing menstrual cycles (Costello and Eden, 2003; Kashyap et al., 2004). A third recent systematic review and meta-analysis comparing metformin with either placebo or no treatment demonstrated that metformin reduced blood pressure, fasting insulin, fasting glucose and serum androgens but there was insufficient evidence of an effect on body weight or hirsutism scores. However, the data for hirsutism scores were derived from a single small trial. The effect of metformin on lipids was variable with evidence of a reduction in serum low density lipoprotein cholesterol level but no demonstrable effect on the levels of total cholesterol, high density lipoprotein cholesterol or triglycerides (Lord et al., 2004). In addition, there is speculation that metformin may reduce long-term consequences of insulin resistance in PCOS women, such as T2DM and CVD (Homburg, 2002).
Therefore, it is important to directly compare the use of OCPs versus metformin in view of the above reservations concerning the appropriateness of OCPs as a therapeutic agent for PCOS and suggestions that metformin may be safer as well as a more effective treatment alternative. The aim of this systematic review is to compare the efficacy and safety of metformin versus OCP in the treatment of women with PCOS.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
The patient population were women with PCOS based on clinical (ovulatory dysfunction, hirsutism, acne, androgen-dependent alopecia), biochemical (hyperandrogenaemia) and/or ultrasound (polycystic ovaries) evidence. Note was taken of whether any of the participants had diabetes mellitus or were taking any other medications which may alter insulin sensitivity. Note was also taken as to whether the subjects of the included studies met the new proposed internationally agreed definition of PCOS (Fauser et al., 2004
). The interventions compared were metformin versus OCP in a direct head-to-head fashion. Primary outcome measures were the clinical parameters of hirsutism, acne, diagnosis of T2DM, CVD (stroke or myocardial infarction) event and endometrial cancer event. Secondary outcome measures were the clinical parameter of menstrual cyclicity (i.e. an initiation of menses or significant shortening of cycles), hormonal parameter of serum total testosterone, metabolic parameters of fasting serum levels of insulin, glucose, total cholesterol and triglycerides and severe (requiring stopping of medication) adverse events. All randomized controlled trials (RCTs) comparing metformin with the OCP were reviewed. Non-RCTs and quasi-randomized trials, in which the method of allocation to different forms of treatment is not truly random, were excluded. Crossover trials were not included unless phase 1 (i.e. pre-crossover) data were available, in which case only these data were used for the purpose of the review.
The literature search aimed to locate RCTs in both English and foreign languages, with no language or other limitations imposed. We searched the Cochrane Menstrual Disorders and Subfertility Group's (MDSG) trials register (September 2005) and the following electronic databases: Cochrane Central Register of Controlled Trials [CENTRAL (Ovid), 3rd Quarter 2005], MEDLINE (1966 to September 2005), CINAHL (1982 to September 2005) and EMBASE (1988 to September 2005). Reference lists of included studies, other relevant review articles and textbooks were checked for additional relevant citations. Pharmaceutical companies were contacted to locate any prospective registered clinical trials. Experts and specialists in the field were also contacted to identify further reported trials.
Study selection
All eligible studies were assessed for relevance to the review objectives and their methodological quality. Study selection was undertaken by two reviewers (M.F.C. and B.S.). Reviewers extracted data independently and assessed whether the studies met the inclusion criteria. Any disagreements were resolved in consensus with J.E. If papers contained insufficient information to make a decision about eligibility, the authors of those papers were contacted in order to seek further information.
The quality of allocation concealment was graded as adequate (A), unclear (B) or inadequate (C), following the detailed descriptions of these categories provided by the MDSG.
Four RCTs comparing metformin with the OCP were identified. (Morin-Papunen et al., 2000, 2003; Harborne et al., 2003b; Rautio et al., 2005) (Figure 1). Three of the four trials comparing metformin with the OCP were conducted by the same investigator centre (Morin-Papunen et al., 2000, 2003; Rautio et al., 2005). The patients in one of these studies (Rautio et al., 2005) included the combined patients of two of the other studies (Morin-Papunen et al., 2000, 2003). The study by Rautio et al. (2005) examined only the metabolic parameters of lipid levels as outcomes, whereas the two studies by Morin-Papunen et al. (2000) (recruited obese women with PCOS) and Morin-Papunen et al. (2003) (recruited non-obese women with PCOS) assessed clinical, hormonal and metabolic (insulin sensitivity and glucose tolerance) parameters.
|
Description and quality assessment of included studies
Table I summarizes the characteristics of the four included RCTs. The diagnosis of PCOS in all four trials required at least two of the following three features: (i) oligomenorrhoea or amenorrhoea, (ii) clinical or biochemical hyperandrogenism and (iii) polycystic ovaries on ultrasound (Morin-Papunen et al., 2000
, 2003; Harborne et al., 2003b; Rautio et al., 2005). Subjects in three (Morin-Papunen et al., 2000, 2003; Rautio et al., 2005) of the four included studies did not meet the new proposed internationally agreed definition of PCOS (Fauser et al., 2004) due to failure to exclude other causes of hyperandrogenism such as hyperprolactinaemia and congenital adrenal hyperplasia.
|
The main inclusion criteria were: PCOS in one of the trials (Rautio et al., 2005
); PCOS whose primary complaint was hirsutism in one of the trials (Harborne et al., 2003b); obese PCOS in one of the trials (Morin-Papunen et al., 2000) and non-obese PCOS in one of the trials (Morin-Papunen et al., 2003). Exclusion criteria included T2DM for all four trials (Morin-Papunen et al., 2000, 2003; Harborne et al., 2003b; Rautio et al., 2005), sex hormones or drugs known to affect carbohydrate metabolism for one of the trials (Harborne et al., 2003b) and sex hormones or drugs known to affect lipid metabolism for all four trials (Morin-Papunen et al., 2000, 2003; Harborne et al., 2003b; Rautio et al., 2005).
The duration of the trials ranged from 4 to 12 months. The dropout rate for the studies ranged from 15% to 44%. The number of withdrawals was similar for both the treatment and control arms in all of the studies. The dose of metformin was 500 mg three times daily for one of the trials (Harborne et al., 2003b) and 500 mg twice daily for the first 3 months increasing to 1000 mg twice daily for the next 3 months for three of the trials (Morin-Papunen et al., 2000, 2003; Rautio et al., 2005). The OCP type was ethinyl estradiol (EE) 35 µg combined with cyproterone acetate (CPA) 2 mg (EE35/CPA2) in all four trials (Morin-Papunen et al., 2000, 2003; Harborne et al., 2003b; Rautio et al., 2005). None of the trials actively changed the subject's lifestyle in terms of diet and exercise.
The included studies randomized a total of 156 (104 excluding duplication of participants as discussed in ‘study selection’ above) women (range for individual trials 20–52). Two (50%) of the four studies randomized fewer than 50 participants. All four trials were graded ‘A—clear’ for allocation concealment performed by on-site computer system utilizing locked files or third party with the random allocation sequence generated by computer generation or random numbers table. All four studies were unblinded and data analysed by available case analyses (trial participants analysed in the groups to which they were randomized and only participants who completed the trial were included) rather than on the preferred intention-to-treat basis (trial participants analysed in the groups to which they were randomized and all participants included). One trial (Rautio et al., 2005) had sample size justification.
Additional information was sought and supplied from all four trials with the lead author responding for two trials (Morin-Papunen et al., 2000, 2003) and the second (Harborne et al., 2003b) and fourth (Rautio et al., 2005) authors for the remaining trials.
Statistical analysis
Statistical analyses were performed in accordance with the guidelines for statistical analysis developed by the Cochrane Collaboration. For dichotomous data, results for each study were expressed as an odds ratio (OR) with 95% confidence intervals (CIs) and combined for meta-analysis with Review Manager (RevMan) software using the fixed-effect model (Peto method). For continuous data, the mean post-treatment/intervention values and standard deviation for each group were measured and the weighted mean differences (WMDs) with 95% CI were calculated. Where different scales measured the same continuous data outcome, the mean post-treatment/intervention values and standard deviation for each group were measured and the standardized mean difference (SMD) with 95% CI was calculated.
Heterogeneity (variations) between the results of different studies was examined by inspecting the scatter in the data points on the graph and the overlap in their CIs and, more formally, by checking the results of the chi-squared tests where P 
0.05 represents statistical homogeneity. P <0.05 or 95% CIs not containing 1.00 (OR) or 0 (WMD) were considered statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
Primary outcome measures
Hirsutism
Three trials comparing metformin versus OCP with a combined total of 69 participants analysed (104 particpants randomized) reported on hirsutism using either Ferriman–Gallway (FG) scoring system (Morin-Papunen et al., 2000
, 2003) or a subjective (patient self-assessed) visual analogue scale from 0 to 10 (Harborne et al., 2003b). Meta-analysis demonstrated no difference in effect on hirsutism between metformin and OCP (SMD –0.18, P = 0.48, 95% CI -0.67 to 0.32) (Figure 2). The SMD statistic is used when pooling continuous data on outcomes measured by different scales (Statistical Analysis). Statistical heterogeneity (variability in the treatment effects being evaluated in the different trials) was present.
|
Acne
There was a single trial comparing metformin versus OCP with 34 participants analysed (52 participants randomized) reporting acne subjectively (patient self-assessed) using a visual analogue scale ranging from 0 to 10 (Harborne et al., 2003b
). This trial demonstrated no significant difference in acne scores between metformin and the OCP (WMD 0.90, P = 0.18, 95% CI -0.40 to 2.20 (Figure 3).
|
Type 2 diabetes mellitus
One trial comparing metformin versus OCP with 18 participants analysed (32 participants randomized) reported diagnosis of T2DM (Morin-Papunen et al., 2000
). No significant difference was seen in the development of T2DM between the metformin and OCP groups (Peto OR 0.17, P = 0.37, 95% CI 0.00 to 8.54) (Figure 4).
|
Cardiovascular disease
There were no trials comparing metformin versus OCP reporting the outcome measure of CVD (stroke or myocardial infarction).
Endometrial cancer
There were no trials comparing metformin versus OCP reporting endometrial cancer as an outcome.
Secondary outcomes measures
Improved menstrual pattern
There were two trials comparing metformin versus OCP with a total of 35 participants analysed (52 participants randomized) reporting on improvement in menstrual cyclicity (Morin-Papunen et al., 2000, 2003). Eighteen of 21 participants on metformin and 20 of 24 participants on the OCP had either oligomenorrhoea or amenorrhoea at baseline. Both trials reported menstrual pattern in terms of days between menses and metformin was significantly less effective in improving menstrual pattern (Peto OR 0.08, P = 0.004, 95% CI 0.01–0.45) (Figure 5).
|
Hormonal: serum total testosterone (nmol l–1)
There were three trials comparing metformin versus OCP with a total of 69 participants analysed (104 participants randomized) reporting on total serum testosterone (Morin-Papunen et al., 2000
, 2003; Harborne et al., 2003b). Meta-analysis demonstrated a significantly higher serum total testosterone with metformin compared with the OCP (WMD 0.54, P = 0.001, 95% CI 0.22–0.86) (Figure 6).
|
Metabolic: fasting insulin (mIU l–1)
There were three trials comparing metformin versus OCP with a total of 69 participants analysed (104 participants randomized) reporting on fasting insulin levels (Morin-Papunen et al., 2000
, 2003; Harborne et al., 2003b). Meta-analysis showed significantly lower fasting insulin levels in favour of metformin (WMD –3.46, P = 0.0005, 95% CI –5.39 to –1.52) (Figure 7). The fasting insulin levels did not change with OCP treatment in any of the three individual trials.
|
Metabolic: fasting glucose (mmol l–1)
Three trials comparing metformin versus OCP with a total of 69 participants analysed (104 participants randomized) reported on fasting glucose levels (Morin-Papunen et al., 2000
, 2003; Harborne et al., 2003b). There was no significant difference in fasting glucose levels between the two interventions (WMD 0.13, P = 0.25, 95% CI -0.09 to 0.35) (Figure 8). The fasting glucose levels did not change with OCP treatment in any of the three individual trials.
|
Metabolic: fasting total cholesterol (mmol l–1)
There were two trials comparing metformin versus OCP with a total of 69 participants analysed (104 participants randomized) reporting on total cholesterol levels (Harborne et al., 2003b
; Rautio et al., 2005). Meta-analysis demonstrated no significant difference in total cholesterol levels between metformin and the OCP (WMD –0.11, P = 0.60, 95% CI –0.53 to 0.30) (Figure 9).
|
Metabolic: fasting triglycerides (mmol l–1)
Two trials comparing metformin versus OCP with a total of 69 participants analysed (104 participants randomized) reported on triglyceride levels (Harborne et al., 2003b
; Rautio et al., 2005). Metformin resulted in a significantly lower triglyceride level than the OCP (WMD –0.48, P = 0.01, 95% CI –0.86 to –0.09) (Figure 10).
|
Severe adverse events (requiring stopping of medication) (gastrointestinal and others)
There were three trials comparing metformin versus OCP with a total of 104 participants randomized and analysed reporting on severe (requiring stopping of medication) adverse events (Morin-Papunen et al., 2000
, 2003; Harborne et al., 2003b). Metformin caused a significantly higher incidence of severe gastrointestinal side effects (i.e. nausea, diarrhoea) (Peto OR 7.75, P = 0.02, 95% CI 1.32–45.71) and a significantly lower incidence of other severe adverse events (i.e. weight gain, high blood pressure, depression, chest pain, headache) (Peto OR 0.11, P = 0.0008, 95% CI 0.03–0.39) compared with the OCP. However, overall, there was no significant difference between metformin and the OCP when taking into account all severe adverse events (Peto OR 0.48, P = 0.18, 95% CI 0.17–1.39) (Figure 11). Not surprisingly, there was significant heterogeneity in the overall severe adverse events meta-analysis due to the different side-effect profiles of metformin and the OCP.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
The OCP is the traditional therapy for the chronic treatment of PCOS, exerting a number of beneficial effects including regularization of menses, amelioration of hirsutism and acne and protection from the development of endometrial cancer (Diamanti-Kandarakis et al., 2003
). However, limited but contradictory observational evidence based on non-randomized studies raises the issue that OCPs may worsen insulin resistance, glucose tolerance and lipid levels in PCOS women and consequently possibly enhance the long-term risk for T2DM and CVD leading to speculation that the insulin-sensitizing agent metformin may be a safer treatment alternative (Diamanti-Kandarakis et al., 2003, Vrbikova and Cibula, 2005). This is the first systematic review and meta-analysis of RCTs comparing metformin versus OCP treatment in women with PCOS. This review has found that up to 12 months treatment with the OCP compared with metformin in PCOS is associated with an improvement in menstrual pattern and serum testosterone levels with the OCP and a reduction in fasting insulin and lower fasting triglyceride levels with metformin. Severe side-effect profiles differ between the two therapies. There were either extremely limited or no data on important clinical outcomes such as the development of T2DM, CVD or endometrial cancer. There was no evidence of a difference in effect between metformin and the OCP on hirsutism, acne and fasting glucose or total cholesterol levels.
The OCP was superior to metformin in improving menstrual pattern with all OCP subjects and 10/16 (62.5%) of metformin subjects reporting such an outcome (Figure 5). This finding was not surprising and consistent with that of a recent systematic review which showed that metformin results in the restoration of regular menses in only 60% of PCOS women with oligomenorrhea or amenorrhea (Costello and Eden, 2003). Unfortunately, there were no trials assessing whether this more favourable OCP effect leads to a reduction in the long-term risk of endometrial cancer compared with metformin.
The hirsutism score data from the three trials were conflicting, thus there is insufficient evidence of relative benefit of either metformin or OCP. Statistical heterogeneity was present with the results for the trial by Harborne et al. (2003b) differing from the trials by Morin-Papunen et al. (2000, 2003) with the former trial demonstrating a significant reduction in hirsutism with metformin, whereas the other two trials showing a trend in the opposite direction (conflicting evidence) (Figure 2). The reasons for this are unclear but may be due to differences in the selection criteria for the PCOS participants, assessment method for hirsutism (described in Results) and the duration of treatment as outlined below.
All 52 participants in the trial by Harborne et al. (2003b) and 18 of 52 participants in the trials by Morin-Papunen et al. (2000, 2003) were clinically hirsute. Hirsutism was assessed by a patient self-assessed visual analogue scale in the Harborne et al. (2003b) trial and by the FG scoring system for the trials by Morin-Papunen et al. (2000, 2003). The duration of treatment was 12 months for Harborne et al. (2003b) and 6 months for Morin-Papunen et al. (2000, 2003) (Table I). There was no evidence of benefit with either metformin or OCP for acne based on 34 women in a single trial (Harborne et al., 2003b) (Figure 3). The OCP was superior to metformin in reducing serum total testosterone levels by 0.5 nmol l–1 (Figure 6). Therefore, the OCP results in a greater reduction in total testosterone levels when compared with metformin, but there are insufficient studies to date assessing whether this more favourable biochemical androgen profile with the OCP leads to a more favourable clinical androgen effect in terms of hirsutism or acne.
The meta-analysis demonstrated significantly lower fasting insulin levels by 3.5 mIU l–1 with metformin treatment when compared with the OCP (Figure 7). All three individual RCTs demonstrated a reduction in fasting insulin levels with metformin and no change in fasting insulin levels with OCP treatment (Morin-Papunen et al., 2000, 2003; Harborne et al., 2003b). There was no evidence of a difference in effect between metformin and the OCP on fasting glucose levels (Figure 8). When one examines the data on fasting glucose levels from the individual three trials, there was no change seen with OCP treatment in all three trials comparing with no change in two (Harborne et al., 2003b; Morin-Papunen et al., 2003) of the three trials and a significant decrease by 0.2 mmol l–1 in the third RCT (Morin-Papunen et al., 2000) with metformin treatment.
The meta-analysis showed no evidence of a difference in effect between metformin and the OCP on fasting total cholesterol levels (Figure 9). Fasting total cholesterol levels did not change with either metformin or OCP therapy in each of the two trials (Harborne et al., 2003b; Rautio et al., 2005). However, a significantly lower fasting triglyceride level by 0.5 mmol l–1 was seen with metformin treatment when compared with the OCP (Figure 10). The fasting triglyceride level did not change in both individual RCTs with metformin treatment. However, OCP treatment resulted in either no change (Harborne et al., 2003b) or an increase (Rautio et al., 2005) in fasting triglyceride levels. In the trial by Rautio et al. (2005), the mean fasting triglyceride levels increased from 1.3 to 1.9 mmol l–1 with OCP treatment. The current recommended upper limit for fasting triglyceride levels in the authors' countries of Australia and New Zealand is 2.0 mmol l–1 (Anonymous, 2001), although other international cut-off levels of 1.7 mmol l–1 do exist (Bloomgarden, 2004).
Overall, the meta-analysis showed no evidence of a difference in severe adverse events requiring stopping of medication between the two treatments (Figure 11). However, metformin caused a significantly higher incidence of severe gastrointestinal side effects (i.e. nausea, diarrhoea), whereas those treated with the OCP experienced a significantly higher incidence of severe other (other than gastrointestinal) adverse events (i.e. weight gain, high blood pressure, depression, chest pain, headache).
|
Limitations |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
There are a number of limitations to this review. The value of any meta-analysis is totally dependent on the quality and lack of bias in its component primary studies. The careful assessment of study validity and heterogeneity of the individual studies is essential in minimizing the risk of inappropriately combining biased and disparate studies leading to misleading systematic reviews (Hughes, 1996
). Although, all the included studies were randomized with adequate allocation concealment to reduce selection bias, none were blinded to outcome assessment or analysed by intention-to-treat analysis, placing these trials at risk of ascertainment and attrition biases, respectively (Guyatt et al., 1993).
Although the inclusion of unpublished studies is controversial (Cook, et al., 1993), reliance upon published studies alone may distort the results of a meta-analysis because positive studies (statistically significant) are more likely to be published than negative ones (non-significant) with the attendant risk for the review to overestimate treatment efficacy (Dickersin et al., 1987). Funnel plot analysis for the identification of publication bias in this review was not performed due to the limited number of studies.
There were a limited number of RCTs (n = 4) comparing metformin versus OCP. A number of results are constrained by small numbers and wide CIs that limit the precision and confidence of conclusions. In addition, the ‘lack of evidence of difference’ for a number of outcomes is not synonymous with ‘evidence of a lack of difference’. Meta-analysis was not possible for a number of important clinical primary outcomes due to either an absence of trials (CVD, endometrial cancer) or the presence of a single trial only (acne, T2DM).
All of the data in this review are from women with PCOS recruited from European centres and three of the four trials were conducted by the same investigator group. This may limit the potential applicability of the results of this review if ethnic variation in baseline risk of adverse outcomes (clinical or metabolic) or response to metformin or the OCP exists. Differences between trial populations, assessment methods, study duration and drug side-effect profiles have resulted in heterogeneity within some of the analyses as discussed above. The mean age of the subjects in the trials ranged from 28 to 31 years, limiting the applicability of the results of this review to not include adolescent women with PCOS. All the trials were of short duration, and therefore, long-term data on the comparison effects between metformin and OCPs are lacking in terms of important clinical outcomes such as hirsutism, acne and development of T2DM, CVD or endometrial cancer.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
From the studies included in this review, there is limited evidence to support that the OCP is more effective than metformin in improving menstrual pattern and reducing serum total testosterone levels. There is also limited evidence to show that metformin is more effective than the OCP in reducing fasting insulin and not increasing triglyceride levels and that metformin causes a higher incidence of severe gastrointestinal adverse effects, whereas the OCP results in a higher incidence of severe non-gastrointestinal adverse effects. However, there is insufficient evidence to show that either metformin or the OCP is more effective in terms of treating hirsutism and acne, preventing the development of T2DM and reducing fasting glucose or total cholesterol levels. There are no data (or evidence) on which to make clinical decisions in terms of development of CVD or endometrial cancer when comparing metformin with the OCP.
The limited RCT evidence to date does not show adverse metabolic risk with the use of the OCP compared with metformin in terms of both clinical (T2DM, CVD) and surrogate (fasting glucose, insulin and total cholesterol levels) metabolic outcomes. However, it should be re-emphasized that all available trials are limited in the duration of metformin or OCP treatment and therefore long-term effects are not known.
The limited number of RCTs comparing metformin versus OCP treatment highlights the need for more long-term, well-designed and executed RCTs directly comparing the efficacy (in terms of both clinical and surrogate hormonal and metabolic outcomes) and safety of these therapies in order to help clarify the preferred long-term medical treatment option for women with PCOS, including adolescents. There is a striking lack of data concerning long-term outcomes, including T2DM, CVD and endometrial cancer, and this should be addressed in longer-term trials.
|
Acknowledgement |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
The authors wish to thank the Cochrane Menstrual Disorders and Subfertility Review Group in Auckland, New Zealand, for their help and support in the conduct and preparation of this review.
|
Footnotes |
|---|
This paper is based on a Cochrane review to be published in The Cochrane Library, 24 January 2007 (see www.thecochranelibrary.com for information) with permission from The Cochrane Collaboration and John Wiley & Sons. Cochrane reviews are regularly updated as new evidence emerges and in response to feedback, and The Cochrane Library should be consulted for the most recent version of the review. The results of a Cochrane review can be interpreted differently, depending on people's perspectives and circumstances. Please consider the conclusions presented carefully. They are the opinions of review authors and are not necessarily shared by The Cochrane Collaboration.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionLimitationsConclusionsAcknowledgementReferences |
|---|
Anonymous. (2001) National Heart Foundation of Australia and The Cardiac Society of Australia and New Zealand. Lipid management guidelines—2001. Med J Aust 175:Suppl_5S57–S88.[Medline]
Baillargeon JP, Iuorno M, Nestler JE. (2003) Insulin sensitizers for polycystic ovary syndrome. Clin Obstet Gynecol 46:325–340.[CrossRef][ISI][Medline]
Balen A. (2004) The pathophysiology of polycystic ovary syndrome: trying to understand PCOS and its endocrinology. Best Pract Res Clin Obstet Gynaecol 18:685–706.[CrossRef][Medline]
Balen A and Michelmore K. (2002) What is polycystic ovary syndrome? Are national views important? Hum Reprod 17:2219–2227.
Bloomgarden ZT. (2004) Dyslipidaemia and the metabolic syndrome. Diabetes Care 27:3009–3016.
Cook DJ, Guyatt GH, Ryan G, Clifton J, Buckingham L, Willan A, McIlroy W, Oxman AD. (1993) Should unpublished data be included in meta-analyses? current convictions and controversies. JAMA 269:2749–2753.[Abstract]
Costello MF and Eden JA. (2003) A systematic review of the reproductive system effects of metformin in patients with polycystic ovary syndrome. Fertil Steril 79:1–13.[CrossRef][ISI][Medline]
Diamanti-Kandarakis E, Kouli CR, Bergiele AT, Filandra FA, Tsianateli TC, Spina GG, Zapanti ED, 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.
Diamanti-Kandarakis E, Baillargeon JP, Iuorno MJ, Jakubowicz DJ, Nestler JE. (2003) A modern medical quandary: polycystic ovary syndrome, insulin resistance, and oral contraceptive pills. J Clin Endocrinol Metab 88:1927–1932.
Dickersin K, Chan S, Chalmers TC, Sacks HS, Smith H Jr. (1987) Publication bias and clinical trials. Control Clin Trials 8:343–353.[CrossRef][ISI][Medline]
Dunaif A, Segal KR, Futterweit W, Dobrjansky A. (1989) Profound peripheral insulin resistance, indepentent of obesity, in polycystic ovary syndrome. Diabetes 38:1165–1174.[Abstract]
Dunn CJ and Peters DH. (1995) Metformin: a review of its pharmacological properties and therapeutic uses in non-insulin dependent diabetes. Drugs 49:721–749.[ISI][Medline]
Ehrmann DA and Rychlik D. (2003) Pharmacologic treatment of polycystic ovary syndrome. Sem Reprod Med 21:277–283.[CrossRef]
Fauser B, Tarlatzis B, Chang J, Azziz R, Legro R, Dewailly D, Franks S, Balen AH, Bouchard P, Dahlgren E, et al. (2004) The Rotterdam ESHRE/ASRM-sponsored PCOS consensus workshop group. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod 19:41–47.
Guyatt GH, Sackett DL, Cook DJ. (1993) Users' guides to the medical literature: II. How to use an article about therapy or prevention: A. are the results of the study valid? JAMA 270:2598–2601.
Harborne L, Flemming R, Lyall H, Norman J, Sattar N. (2003a) Descriptive review of the evidence for the use of metformin in polycystic ovary syndrome. Lancet 361:1894–1901.[CrossRef][ISI][Medline]
Harborne L, Fleming R, Lyall H, Sattar N, Norman J. (2003b) Metformin or antiandrogen in the treatment of hirsutism in polycystic ovary syndrome. J Clin Endocrinol Metab 88:4116–4123.
Hardiman P, Pillay OC, Atiomo W. (2003) Polycystic ovary syndrome and endometrial carcinoma. Lancet 361:1810–1812.[CrossRef][ISI][Medline]
Homburg R. (1996) Polycystic ovary syndrome—from gynaecological curiosity to multisystem endocrinopathy. Hum Reprod 11:29–39.
Homburg R. (2002) Should patients with polycystic ovary syndrome be treated with metformin. Hum Reprod 17:853–865.
Hughes EG. (1996) Systematic literature review and meta-analysis. Semin Reprod Endoc 14:161–169.
Kashyap S, Wells GA, Rosenwaks Z. (2004) Insulin-sensitizing agents as primary therapy for patients with polycystic ovarian syndrome. Hum Reprod 19:2474–2483.
Knochenhauer ES, Key TJ, Kahsar-Miller M, Waggoner W, Boots LR, 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.
Lord JM, Flight IHK, Norman RJ. (2003) Insulin-sensitising drugs (metformin, troglitazone, rosiglitazone, pioglitazone, D-chiro-inositol) for polycystic ovary syndrome. The Cochrane Database of Systematic Reviews 2 DOI: 10.1002/14651858. CD003053.
Morin-Papunen LC, Vauhkonen I, Koivunen RM, Ruokonen A, Martikainen HK, Tapanainen JS. (2000) Endocrine and metabolic effects of metformin versus ethinyl estradiol–cyproterone acetate in obese women with polycystic ovary syndrome: a randomized study. J Clin Endocrinol Metab 85:3161–3168.
Morin-Papunen L, Vauhkonen I, Koivunen R, Ruokonen A, Martikainen H, Tapanainen JS. (2003) Metformin versus ethinyl estradiol-cyproterone acetate in the treatment of nonobese women with polycystic ovary syndrome: a randomized study. J Clin Endocrinol Metab 88:148–156.
Nestler JE, Stovall D, Akhter N, Iuorno MJ, Jakubowicz MJ. (2002) Strategies for the use of insulin-sensitizing drugs to treat infertility in women with polycystic ovary syndrome. Fertil Steril 77:209–215.[CrossRef][ISI][Medline]
Ovalle F and Azziz R. (2002) Insulin resistance, polycystic ovary syndrome, and type 2 diabetes mellitus. Fertil Steril 77:1095–1105.[CrossRef][ISI][Medline]
Rautio K, Tapanainen JS, Ruokonen A, Morin-Papunen LC. (2005) Effects of metformin and ethinyl estradiol-cyproterone acetate on lipid levels in obese and non-obese women with polycystic ovary syndrome. Eur J Endocrinol 152:269–275.
Utiger RD. (1996) Insulin and the polycystic ovary syndrome. New Engl J Med 335:657–658.
Vrbikova J and Cibula D. (2005) Combined oral contraceptives in the treatment of polycystic ovary syndrome. Hum Reprod Update 11:277–291.
Wild RA. (2002) Long-term health consequences of PCOS. Hum Reprod Update 8:231–241.
Submitted on November 29, 2006; accepted on January 3, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Jensterle, A. Janez, B. Mlinar, J. Marc, J. Prezelj, and M. Pfeifer Impact of metformin and rosiglitazone treatment on glucose transporter 4 mRNA expression in women with polycystic ovary syndrome. Eur. J. Endocrinol., June 1, 2008; 158(6): 793 - 801. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Cosma, B. A. Swiglo, D. N. Flynn, D. M. Kurtz, M. L. LaBella, R. J. Mullan, M. B. Elamin, P. J. Erwin, and V. M. Montori Insulin Sensitizers for the Treatment of Hirsutism: A Systematic Review and Metaanalyses of Randomized Controlled Trials J. Clin. Endocrinol. Metab., April 1, 2008; 93(4): 1135 - 1142. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||