Hum. Reprod. Advance Access originally published online on September 30, 2005
Human Reproduction 2006 21(1):80-89; doi:10.1093/humrep/dei311
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Combined lifestyle modification and metformin in obese patients with polycystic ovary syndrome. A randomized, placebo-controlled, double-blind multicentre study
1 Department of Reproductive Medicine, Clarendon Wing, The General Infirmary, Leeds LS2 9NS, UK 2 Department of Reproductive Medicine, St Mary’s Hospital, London W2 and 3 Department of Clinical Biochemistry, The General Infirmary, Leeds LS1 3EX, UK
4 To whom correspondence should be addressed. E-mail: Adam.balen{at}leedsth.nhs.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
BACKGROUND: It has been reported that women with polycystic ovary syndrome (PCOS) benefit from metformin therapy. METHODS: A randomized, placebo-controlled, double-blind study of obese (body mass index >30 kg/m2), oligo-/amenorrhoeic women with PCOS. Metformin (850 mg) twice daily was compared with placebo over 6 months. All received the same advice from a dietitian. The primary outcome measures were: (i) change in menstrual cycle; (ii) change in arthropometric measurements; and (iii) changes in the endocrine parameters, insulin sensitivity and lipid profile. RESULTS: A total of 143 subjects was randomized [metformin (MET) = 69; placebo (PL) = 74]. Both groups showed significant improvements in menstrual frequency [median increase (MET = 1, P < 0.001; PL = 1, P < 0.001)] and weight loss [mean (kg) (MET = 2.84; P < 0.001 and PL = 1.46; P = 0.011)]. However, there were no significant differences between the groups. Logistic regression analysis was used to analyse the independent variables (metformin, percentage of weight loss, initial BMI and age) in order to predict the improvement of menses. Only the percentage weight loss correlated with an improvement in menses (regression coefficient = 0.199, P = 0.047, odds ratio = 1.126, 95% CI 1.001, 1.266). There were no significant changes in insulin sensitivity or lipid profiles in either of the groups. Those who received metformin achieved a significant reduction in waist circumference and free androgen index. CONCLUSIONS: Metformin does not improve weight loss or menstrual frequency in obese patients with PCOS. Weight loss alone through lifestyle changes improves menstrual frequency.
Key words: menstrual frequency/metformin/obese/polycystic ovary syndrome/weight loss
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The polycystic ovary syndrome (PCOS) is the commonest endocrine disturbance in women (Balen and Michelmore, 2002
) and the commonest cause of anovulatory infertility. PCOS is a heterogeneous disorder with features including hyperandrogenism, menstrual irregularity and obesity (Balen et al., 1999; Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group, 2004). The association between insulin resistance, compensatory hyperinsulinaemia and hyperandrogenism have provided insight into the pathogenesis of PCOS (Tsilchorozidou et al., 2004). Insulin resistance occurs in both slim and overweight women with PCOS, although there is debate on the proportion of women with PCOS with reduced insulin sensitivity (Cibula et al., 2004). At least 40% of women with PCOS are obese (Balen et al., 1995) and they are more insulin resistant than weight-matched individuals with normal ovaries (Dunaif et al., 1995; Morales et al., 1996).
Obesity and particularly abdominal obesity as indicated by an increased waist:hip ratio is correlated with reduced fecundity (Zaadstra et al., 1993; Kirchengast and Huber, 2004; Pasquali et al., 2003), menstrual disorders and hyperinsulinaemia (Conway et al., 1990; Lord and Wille, 2002). Obesity correlates with an increased rate of menstrual cycle disturbance and infertility (Kiddy et al., 1990; Balen et al., 1995). Weight loss improves the endocrine profile, the menstrual cyclicity, the likelihood of ovulation and of a healthy pregnancy (Pasquali et al., 1989; Kiddy et al., 1992; Huber-Buchholz et al., 1999). Studies by Clark et al. (1995, 1998), demonstrated that weight loss achieved by an exercise schedule, combined with a hypo-caloric diet over a 6 month period, improved insulin sensitivity, endocrine parameters, menstrual regularity, the frequency of spontaneous ovulation and the chance of pregnancy.
Even a modest weight loss of 2–5% of total body weight can restore ovulation in overweight women with PCOS as well as achieving a reduction of central fat and an improvement in insulin sensitivity (Huber-Buchholz et al., 1999). Rather than absolute weight, it is the distribution of fat that is important with android (central) obesity being more of a risk factor than gynaecoid obesity (Despres et al., 2001; Lord and Wille, 2002). Visceral adipose tissue is more metabolically active than subcutaneous fat and the amount of visceral fat correlates with insulin resistance and hyperinsulinaemia. Weight reduction of 5–10% may result in 
30% loss of visceral adipose tissue (Despres et al., 2001) and this may explain why a modest weight loss can significantly improve metabolic and reproductive function. Waist circumference has been shown to correlate better with visceral fat than waist:hip ratio (WHR) (Lord and Wille, 2002), and a waist circumference in women >88 cm is indicative of an increased metabolic risk (Despres et al., 2001).
Lifestyle modification is a key component for the improvement of reproductive function for overweight, anovulatory women with PCOS (Norman et al., 2002, 2004; Pasquali et al., 2003). Weight loss should therefore be encouraged prior to ovulation induction treatments, since these are less effective when the body mass index (BMI) is >28–30 kg/m2 (Hamilton-Fairley et al., 1992). Monitoring treatment is also harder in the obese as visualization of the ovaries is more difficult which raises the risk of multiple ovulation and multiple pregnancy. Furthermore, pregnancy carries greater risks in the obese, for example: miscarriage, gestational diabetes, hypertension and problems with delivery (Gjonnaess et al., 1989; Sebire et al., 2001; Cedergren, 2004; Linné, 2004).
It is logical to assume that therapy that achieves a fall in serum insulin concentrations should improve the symptoms of PCOS (Norman et al., 2004). The biguanide metformin both inhibits the production of hepatic glucose, thereby decreasing insulin secretion, and enhances insulin sensitivity at the cellular level (Matthaei et al., 2000). The efficacy of metformin in PCOS was first described by Velazquez et al. (1994) and a number of small, and often short duration, observational studies followed which showed variable outcomes. Most of the randomized studies have involved only a small number of participants. Indeed in a systematic review by Costello and Eden (2003), nine out 12 published studies on the effects of metformin alone on the menstrual cycle in women with PCOS had a sample size of <30 women. Lord et al. (2003) published a systematic review in the Cochrane Database which concluded that metformin has a beneficial effect for women with PCOS, by reducing serum insulin concentrations and thereby lowering androgen levels and improving reproductive outcomes. Back in 1997 we conceived what we anticipated to be an appropriately powered, prospective randomized, double-blind, placebo-controlled multicentre study to evaluate the combined effects of lifestyle modification and metformin on obese anovulatory women (BMI >30 kg/m2) with PCOS. The study has taken a considerable time to complete and here we present our findings.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Women were recruited from infertility clinics with anovulatory PCOS and a BMI of >30 kg/m2, aged between 18 and 39 years inclusive and a desire to conceive. Anovulation was defined as the presence of amenorrhoea or oligomenorrhoea (cycle length >35 days) (Munster et al., 1993
; Berek et al., 1996) and the absence of ovarian follicular activity on serial ultrasound scans. The patients had not received ovulation induction therapy from the fertility clinic as the usual criterion for any form of ovulation induction (clomiphene citrate or gonadotropin therapy) was a BMI of <30 kg/m2.
PCOS was defined as the presence of polycystic ovaries on transvaginal scan, >10 cysts, 2–8 mm in diameter, usually combined with increased ovarian volume >10 cm3, and an echo-dense stroma (after the transabdominal ultrasound criteria of Adams et al., 1985), together with either oligomenorrhoea or amenorrhoea. Many patients also had clinical or biochemical hyperandrogenism, although this was not an entry criterion for the study. All patients had a baseline androgen profile, including measurement of testosterone and androstenedione. If either were significantly elevated, additional tests were performed to exclude hypercortisolism and congenital adrenal hyperplasia (CAH) (full steroid profile, 24 h urinary cortisol and adrenocorticotrophin hormone stimulation test). When the study was devised the Rotterdam consensus definitions of PCOS (2004) and of the polycystic ovary (Balen et al., 2003) had not been defined, although all of our patients would have been classified as having PCOS by those criteria.
Pretreatment inclusion criteria also included the presence at least one patent Fallopian tube and a normal semen analysis from the male partner. All participants had normal serum prolactin concentrations, thyroid, renal and liver function and haematological indices, including serum B12 concentrations.
Exclusion criteria included concurrent hormone therapy within the previous 6 weeks, any chronic disease that could interfere with the absorption, distribution, metabolism or excretion of metformin, and renal or liver disease. Patients with significant systemic disease or diabetes (Type 1 or 2) were excluded. Patients with irregular menstrual bleeding were thoroughly assessed to exclude pathology of the genital tract other than PCOS and a negative pregnancy test was a prerequisite for commencing treatment.
Protocol
A multicentre research ethics committee approval (MREC 1999/8/12) and the local research ethics committee approval of each participating centre were obtained. After obtaining written consent, a full physical examination was performed including assessment of BMI, waist and hip circumference and blood pressure. A baseline transvaginal ultrasound scan was performed to assess ovarian morphology, uterine size and endometrial thickness. A standardized 75 g oral glucose tolerance test (OGTT) was performed with measurement of fasting insulin concentration and glucose at 0 and 120 min. Baseline serum endocrinology included the measurement of FSH, LH, testosterone, sex hormone-binding globulin (SHBG), total cholesterol and triglycerides.
The subjects were randomized to receive either metformin or placebo. The randomization process was carried out by the clinical trials office in the pharmacy department and blinded to patients and investigators. A block-of-four randomization technique was performed using random tables from Linder et al. (1970). The code was kept in the trial office until the last patient completed the study. Placebo tablets for metformin were identical in appearance (size and colour) to metformin and were supplied by Penn Pharmaceuticals Ltd (Tredegar, Gwent). One tablet (metformin 850 mg or placebo) was prescribed to be taken 12 hourly for a period of 6 months.
Patients in each group received standardized dietary advice from a research dietitian. Each subject was assessed by the dietitian and an individualized diet [high in carbohydrate (50%) and low in fat (10%)] was given with the aim of a reduction in daily intake by 500 kcal. Written information was given on PCOS and appropriate information on a balanced weight-reducing diet. The patients were also encouraged to increase daily exercise (such as walking, using stairs) by 15 min, although this was not formally assessed. The participants received further encouragement to adhere to the regime at the monthly review visits.
Each participant was assessed monthly with a re-evaluation of anthropometric measurements, endocrine and biochemical parameters together with an ultrasound scan and record of the patient’s menstrual cycle. Side-effects of the treatment and reason for any withdrawals from the study were recorded. The assessment was performed by the same person in each centre (usually the research nurse). All nursing and medical personnel were blind to the treatment arm, with the research pharmacy in Leeds being the only place where this information was held for the duration of the study. Compliance was assessed by the return of empty drug containers.
Outcome measures
The primary outcome measures were: (i) change in menstrual cycle; (ii) change in arthropometric measurements; and (iii) changes in the endocrine parameters, insulin sensitivity and lipid profile. The main secondary outcome measure was pregnancy rate.
Power calculation
In the study by Velazquez et al. (1997a), metformin alone improved menstrual regularity in 21/40 (53%) of subjects. If we anticipate an overall 83% improvement with a combination of diet and metformin (i.e. a further 30% improvement compared with metformin alone), the standardized difference (d) would be 0.64. The chosen power in the study was 90% with a type I error of 0.05. From the power table (Machin and Campbell, 1987), when d = 0.64 and the power = 0.90, the projected sample size was 110, with 55 subjects in each arm of the study. When the study was designed the literature from which to calculate power was limited, if we were to consider the more recently published Cochrane meta-analysis by Lord et al. (2003), which reported an overall improvement in ovulation rates in 71/156 and 37/154 subjects in the metformin group and the control group respectively with an odds ratio of 3.88 (95% CI 2.25, 6.69) versus placebo for rate of ovulation in favour of metformin. Based on these recent values, the standardized difference is 0.57 with the projected sample size of 130. At the end of the study period, the actual recruitment exceeded this value (see below).
Biochemical assays
All the samples were stored at –20°C and were analysed in the biochemistry department of the coordinating centre. The analyses were as previously described (Wijeyaratne et al., 2002). Plasma glucose was measured using an enzymatic colorimetric assay (Hitachi, Roche) with intra-assay coefficients of variation (CV) of 1.9% at 20.2 mmol/l and 30% at 2.4 mmol/l. A time-resolved fluoroimmunoassay (AutoDELFIA; Perkin Elmer) was used to measure insulin and serum SHBG concentrations, with plasma insulin intra-assay CV of 1.7% at 180.96 pmol/l and 2.4% at 33.9 pmol/l and inter-assay CV of 3.5% at 180.96 pmol/l and 2.3% at 33.9 pmol/l. The serum SHBG intra-assay CV was 6% at 103.36 nmol/l and 7% at 14.88 nmol/l; and the inter-assay CV was 1% at 103.36 nmol/l and 1% at 14.88 nmol/l. Serum testosterone was measured after organic extraction using an in-house radioimmunoassay with an inter-assay CV of 7.7% at 2.20 nmol/l. Free androgen index (FAI) was derived from the ratio of the total testosterone concentration (nmol/l) to the concentration of SHBG (nmol/l) x 100.
Data analysis and statistics
The insulin sensitivity (IS) was calculated from the Quantitative Insulin Sensitivity Check Index (QUICKI), described by Katz et al. (2000). QUICKI = 1/[log(I0) + log(G0)], with I0 = fasting insulin concentrations in mIU/ml (conversion from pmol/l to mIU/ml: multiplied by a factor of 0.144) and G0 = fasting glucose concentrations in mg/dl (conversion from mmol/l to mg/l: multiplied by a factor of 18.0).
The hyperinsulinaemic–euglycaemic glucose clamp technique is the ‘gold standard’ for quantifying insulin sensitivity in vivo because it directly measures the effects of insulin to promote glucose utilization under steady state conditions; an alternative reference method is the intravenous glucose tolerance test (IV-GTT). Both require sophisticated investigation centres, are labour intensive, expensive and cannot really be performed for large scale studies. In routine clinical practice an OGTT or simple ratios of fasting glucose/insulin are fairly sensitive (Moran and Norman, 2004). More accurate indices of insulin sensitivity and secretion derived from fasting plasma insulin and blood glucose concentrations are reasonable substitutes for the euglycaemic clamp and IV-GTT, these include the HOMA and QUICKI methods (Hanson et al., 2000).
The homeostatic model assessment (HOMA) is a computer-generated model consisting of a series of non-linear empirical equations solved numerically to predict glucose, insulin and C-peptide concentrations in the fasting state for the assessment of pancreatic beta cell function and insulin sensitivity (Matthews et al., 1985). The use of HOMA correlates well with the euglycaemic clamp method and the IV-GTT but cannot be compared between different studies unless the insulin assay is standardized (Bonara et al., 2000). The estimation of the QUICKI provides a robust and reproducible estimate of insulin sensitivity that shows excellent linear correlation with the gold standard clamp estimation with similar variability and discrimination power (Katz et al., 2000). The relative advantages of QUICKI over HOMA include the fact that the data derived from a single blood sample performs just as well as an average of multiple sampling, and the simplicity of the mathematical model. Furthermore, therapeutic changes in insulin sensitivity have been as readily demonstrated with this simple method as with the euglycaemic clamp (Mathur et al., 2001). A recent review on the determination of insulin sensitivity in PCOS has highlighted the good correlation of QUICKI with the clamp technique (Kauffman and Castracane, 2003).
Data were analysed on the basis of intention to treat. All the subjects who withdrew within the first 4 months of the study period, excluding those who conceived, were classified as non-responders. This is because we wished to include only those who completed 
4 months, and preferably 6 months of the trial. For parametric data, the assumption of normal distribution was assessed by a normal plot and the Kolmogorov–Smirnov test. The assumption of the two groups having the same variances was tested by using the F-test. Paired t-test or two-sample t-test was applied as indicated. When the data did not meet the above assumptions, a log10 transformation of the data was carried out. If the transformed data was still not meeting the assumptions, non-parametric methods, Wilcoxon signed rank test or Mann–Whitney test were applied. P < 0.05 was considered to be statistically significant. The Z-test was used to analyse the two proportions with Yates’ correction.
In the multiple linear regression analysis, the same normality test was used as in the t-test and the test for constant variance was computed by using the Spearman rank correlation between the absolute values of the residuals and the observed value of the dependent variable. When the criteria of normality or constant variance were not met, a log10 transformation of the data was performed. Durbin–Watson statistic was used to test residuals for their independence of each other.
In the logistic regression analysis, the regression coefficients computed by minimizing the sum of squared residuals in multiple logistic regression are also the maximum likelihood estimates. P is the P-value calculated for the Wald statistic, which is the regression coefficient divided by the SE. All the statistical analyses were performed using SigmaStat, version 2.
Recruitment progress
During a 4 year period, between 1999 and 2003, a total of eight centres took part in the recruitment process. A total of 183 women were screened for inclusion in the study. Of these, 40 women were excluded due to previously undiagnosed tubal disease or co-existing male factor infertility. As a result, a total of 143 subjects were randomized to receive metformin (n = 69) or to receive placebo (n = 74) (Figure 1). In the metformin arm, 13 subjects withdrew within the first 4 months of the trial (11 due to side-effects and two due to spontaneous pregnancies). Eight women withdrew from the placebo arm (six due to ‘side-effects’ and two due to spontaneous pregnancies, within the first 2 months of the study). The difference in the drop-out rates, excluding because of pregnancy (metformin, 15.9% versus placebo, 8.0%) was not significant (P = 0.229, 95% CI –2.69 to 18.5). At the end of the study, the numbers of patients who completed the trial in the metformin and placebo arms were 56 and 66 respectively. Compliance was high and the drop-out rate relatively low as these were patients motivated by a desire to conceive and the knowledge that they needed to attain a BMI of <30 kg/m2 to qualify for ovulation induction.
|
The total number of patients per centre, with those who withdrew in parentheses, were: Leeds 65 (6), St Mary’s Hospital, London 41 (1), MRC Reproductive Medicine Centre, Edinburgh 3 (1), Royal Shrewsbury Hospital, Shrewsbury 12 (4), Royal Free Hospital, London 6 (0), St Bartholomew’s Hospital, London 4 (3), Hope Hospital, Salford 4 (0) and The Jessop Hospital for Women, Sheffield 8 (2).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Demographic data
There were no significant differences in the baseline characteristics of the subjects between the two groups (Table I). In the metformin and placebo groups respectively, the mean BMI (37.6 versus 38.9 kg/m2), the median number of menstrual cycles in the preceding 6 months (2 versus 2), the mean waist circumference (111.9 versus 108.8 cm) and waist:hip ratio (WHR) (0.907 versus 0.900) were similar. The anthropometric measurements of the subjects who withdrew prematurely were also not significantly different from those who completed the study (data not shown).
|
As expected, there was a positive correlation between insulin sensitivity and serum SHBG concentrations (log insulin sensitivity = –0.593 + 0.093 x (log SHBG), adjusted R2 = 0.11, P = 0.001). Additionally, there was a negative correlation between insulin sensitivity and serum triglyceride concentrations and BMI (log insulin sensitivity = –0.454 – 0.061 x (log triglycerides), adjusted R2 = 0.060, P = 0.011), even after adjustment for age, waist circumference and serum testosterone concentration. Surprisingly, no association between waist circumference and serum insulin concentration and insulin sensitivity was observed.
The mean duration of infertility was similar in each group [MET 4.5 (SD 2.9) years versus PLA 4.9 (3.0) years, P = 0.624]. There was no difference in the percentage of primary infertility (MET 69% versus PLA 73%, P = 0.851) or subjects who had previously been prescribed clomiphene citrate, usually by their primary care physician and not in the context of the fertility clinic, where body mass would have precluded treatment (MET 43% versus PLA 49%, P = 0.718).
Menstrual frequency
At the end of the study period, the menstrual cycles over the time-course of the study increased significantly with a median of improvement of one menstrual cycle per 6 months in both groups (Tables II and III). However, there was no difference between the groups (P = 0.580). Patients who menstruated <4 weeks from starting treatment were not considered to have ovulated in response to the study. A number of studies have used menstrual frequency as an assessment of reproductive function women with PCOS and an improvement in menstrual regularity is considered to be a good surrogate for ovarian function and ovulatory frequency in women with PCOS (Morin-Papunen et al., 1998; Fleming et al., 2002; Haas et al., 2003). Furthermore, Kolstad et al. (1999) studied the relationship between menstrual cycle pattern and fertility. Thus the observed improvement in menstrual frequency can be viewed as an indication of improvement of ovulation rate and potential fecundity.
|
|
On the basis of intention to treat (ITT), 36 women (52.2%) in the metformin group and 43 women (58.1%) in the placebo group experienced improvement in menses. However, the difference between the two groups was not significant (P = 0.589, 95% CI –10.4, 22.2).
Anthropometric measurements
Significant reductions in body weight and BMI were observed in both groups (Tables II and III). However, the changes in the means between the groups were not significant (–1.02 versus –0.46, 95% CI –1.15, 0.03, P = 0.063). The study was not powered to determine a difference in weight even though the metformin group lost twice as much weight as the placebo group. There was a significant reduction of waist circumference in the metformin group (before 113.5 cm, after 111.1 cm, P = 0.002) (Table II) but not in the placebo group (before 108.5 cm, after 109.1 cm, P = 0.764) (Table III). The difference in the changes of the mean values between the two groups was not statistically significant (–2.34 versus +0.58, 95% CI –7.14, 1.30, P = 0.173). Similarly, the changes in the mean of both systolic and diastolic blood pressure were not significantly different between the two groups.
Endocrine parameters and lipid profiles
Both the fasting insulin and glucose data were skewed and therefore logarithmic transformations were performed on the data before analysis. The geometric means of the fasting insulin concentrations in the metformin group did not change significantly over the course of the study (baseline 72.8 pmol/l, final 80.7 pmol/l, ratio of means 1.11, P = 0.524, Table II). Similarly, no significant changes in the geometric means of the fasting insulin concentrations in the placebo group occurred (baseline 74.1 pmol/l, final 81.8 pmol/l, ratio of means 1.10, P = 0.438, Table III). The difference between the changes between the two treatment arms was also not significantly different (1.11 versus 1.10, 95% CI 0.672–1.49, P = 0.985). Similarly, there were no significant changes in fasting glucose concentrations within and between groups (Tables II and III). Improvements in insulin sensitivity were not observed in either the metformin group or the placebo group (Tables II and III). The changes of means in insulin sensitivity were also not different between the two groups (data not shown).
There were no significant changes in the geometric mean SHBG concentrations in either the metformin or placebo arms (Tables II and III), neither was there a difference between the groups (data not shown). There was, however, a significant reduction in the FAI in the metformin arm of the study, with a mean ratio (final:baseline) of 0.84 (95% CI 0.73, 0.96, P = 0.013) and this was because of a significant fall in total testosterone of –0.3 nmol/l (95% CI –0.08, –0.47, P = 0.008) (Table II). This was confirmed by multiple linear regression analysis after adjustment for baseline BMI, change in insulin sensitivity and the percentage of weight change (P = 0.046, Table IV).
|
At the end of the study period, both the total cholesterol and triglyceride concentrations remained unchanged (Tables II and III) with no between-group differences (data not shown).
Pregnancy rates
There were two pregnancies in each arm of the study within 2 months of commencing and a further four pregnancies in the metformin arm in the 5th and 6th months of the study. The total numbers of conceptions in the metformin (8.7%) and the placebo (2.7%) groups were not significantly different (P = 0.233, 95% CI –1.5, 13.5). Based on our findings, the standardized difference is 0.26 and the required sample size to assess a difference in pregnancy rates would be 600 subjects.
Subgroup analysis of those who lost weight
On the basis of intention to treat (ITT), 36 women (52.2%) in the metformin group and 43 women (58.1%) in the placebo group experienced improvement in menses. However, the difference between the two groups was not significant (P = 0.589, 95% CI –10.4, 22.2). If these data are analysed by completion of protocol, the difference is still not significant (P = 0.94, 95% CI –18, 16).
Forty-two subjects (60.8%) in the metformin group and 35 subjects (47.3%) in the placebo group managed to lose weight. The difference between the two groups was not significant (P = 0.147, 95% CI –28.5, 29.9). When we calculated the actual percentage weight change (PWC) [100% x (baseline weight – end of study weight)/baseline weight] among only those women who managed to lose weight, we showed that the mean percentage of weight loss in the metformin and placebo groups was 3.98 and 4.41% respectively. The difference was not significant (P = 0.554, 95% CI –1.88, 1.02).
By using multiple logistic regression analyses of the improvement in menses on the PWC, the use of metformin, the baseline BMI and age, we were able to demonstrate that weight loss (a positive value of PWC) had a significantly positive effect on improvement in menses (P = 0.047, regression coefficient = 0.199, odds ratio 1.126, 95% CI 1.00, 1.27). The use of metformin had no influence on menstrual frequency in our study population.
The best model to predict the improvement in menses is 0.127 x (PWC) + 0.098 x (initial BMI) – 3.185 (see Table V). This implies that the greater the BMI the more likely it was that improvement in menses would have been experienced through weight loss.
|
Analysis of those with the metabolic syndrome
The metabolic syndrome is defined as requiring three out of the following five criteria: waist circumference >88 cm, elevated triglycerides 1.7 mmol/l, lowered high-density lipoprotein cholesterol <1.3 mmol/l, elevated blood pressure (130/85 mmHg) and impaired glucose tolerance test. Twenty-six of those in the metformin arm and 23 in the placebo arm had the metabolic syndrome. There was no difference in outcome between the metformin group and placebo group respectively in the median change of menstrual frequency (1 versus 1, P = 0.916), percentage weight loss (3.14 versus 2.65% P = 0.79), change in waist circumference (–1.5 versus –0.93 cm, P = 0.692), change in serum testosterone concentration (0.889 versus 0.968 nmol/l, P = 0.408), change in FAI (0.891 versus 0.995, P = 0.435), change in insulin sensitivity (–0.003 versus 0.000, P = 0.914) or either cholesterol or triglyceride concentrations.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
We report a large randomized controlled trial (RCT) to investigate the effects of metformin on very obese patients with anovulatory PCOS. The duration of the study period (6 months) and the dose of metformin used (850 mg, twice daily) were the longest and the highest of the RCT reported in the Cochrane database (Lord et al., 2003
). We were unable to demonstrate that metformin had an additional benefit on the improvement of menstrual frequency over weight loss through lifestyle modification and, furthermore, in the study population metformin did not induce weight loss. After adjustment for baseline BMI and age, only weight loss, but not the use of metformin, was associated with a significant improvement in menstrual cyclicity. In addition, the higher the BMI, the more likely women with PCOS were to benefit from weight loss with respect to improvement of menstrual frequency.
The entry criteria required BMI to be >30 kg/m2, yet the mean BMI was 38 kg/m2 and comprised typical central obesity. These were patients who would not be suitable for ovulation induction for anovulatory infertility because of their obesity and so had not yet been enrolled in the ovulation induction programme, although some had previously received clomiphene citrate from their primary care physician before referral to the fertility clinic. The rate of withdrawal in the metformin group was not significantly different from the placebo group and was lower than that reported by Fleming et al. (2002) in their large trial in which 42% dropped out of the metformin arm compared with 17% of the placebo arm. This may be explained by the fact that all of our patients had a wish to conceive and may therefore have had a greater incentive to adhere to the protocol.
A surprise finding was the lack of change in insulin sensitivity in either the metformin or placebo groups. This is probably explained by the extreme obesity of our patients and the relatively small amount of weight lost. It has been demonstrated that insulin sensitivity and androgen concentrations are unlikely to improve in patients who lose <5% of their initial weight (Kiddy et al., 1992). Furthermore, the effect of metformin in women with PCOS is reduced by increasing obesity (Crave et al., 1995; Fleming et al., 2002; Maciel et al., 2004). Our findings were similar to the study of Ehrmann et al. (1997) in which the average BMI was 39 kg/m2. Furthermore, the dose of metformin (850 mg twice daily) may be insufficient in this group of patients and we are currently performing a dose-finding study, using different doses at different body weights.
Metformin, however, did improve the FAI, secondary to a significant fall in total testosterone without a change in the insulin sensitivity or SHBG. This observation suggests that metformin may have a direct effect on ovarian steroidogenesis without effecting a change in circulating insulin concentrations (Pirwany et al., 1999; la Marca et al., 2002; Mansfield et al., 2003). There is a consensus that metformin has an additive effect in achieving ovulation and pregnancy when combined with drugs to induce ovulation (mainly clomiphene citrate) (Costello and Eden, 2003; Lord et al., 2003). The effect may be quick and this too supports the possibility of a direct effect on the ovary rather than a systemic effect on metabolism.
The use of metformin and other insulin-lowering or -sensitizing agents has excited much interest in the management of PCOS. The literature is replete with studies of varying design, using varying regimens and assessing different outcomes. A relatively small number of these studies (a total of 13) have been of appropriate design to be included in the Cochrane systematic review (Lord et al., 2003). This included seven studies comparing metformin with placebo in a total of 310 patients, which showed that metformin was beneficial for ovulation (odds ratio 3.88, 95% CI 2.25, 6.69, P < 0.0001). The largest study to be included in this series was of 92 patients (Fleming et al., 2002). The meta-analysis also demonstrated that metformin was effective in reducing fasting insulin and total testosterone concentrations but had no effect on BMI or waist circumference (Lord et al., 2003).
Costello and Eden (2003) in their systematic review reached similar conclusions and again a wide range of entry criteria were reported. In particular the average ‘mean BMI’ of those studies that compared metformin with placebo was 31.3 kg/m2 (range 21.4–39.8 kg/m2). There were variable effects reported, with not all studies demonstrating an improvement in insulin sensitivity or fall in testosterone levels (Costello and Eden, 2003). As with our study, neither of the two RCT that reported an improvement in menstrual cyclicity showed a fall in BMI; both reported a fall in testosterone concentrations and only one an improvement in fasting insulin (Moghetti et al., 2000; Pasquali et al., 2000).
Pasquali et al. (2000) studied 20 obese women with PCOS with a control group of 20 obese women without PCOS who were comparable for age and pattern of body fat distribution. All were given a low-calorie diet (1200–1400 kcal/day) for 1 month, after which they were randomized to receive metformin (850 mg twice daily) or placebo for 6 months. Metformin treatment reduced body weight and BMI significantly more than placebo in both PCOS and control women. Fasting insulin decreased significantly in both PCOS women and controls and testosterone concentrations decreased only in PCOS women treated with metformin. SHBG concentrations remained unchanged in all PCOS women, although in the control group, they significantly increased after both metformin and placebo (Pasquali et al., 2000). Thus once again the effects of metformin appear to vary in different study populations.
Women with anovulatory PCOS who lose weight experience an improvement in ovarian function, ovulation and anthropometric indices (Clark et al., 1995; Crosignani et al., 2003). The key component of diet should be calorie restriction, rather than the composition of the diet itself (Moran et al., 2003; Stamets et al., 2004). A recent study randomized 38 women with a mean BMI of >39 kg/m2 to receive either advice on lifestyle modification (aiming for 500–1000 calorie deficit per day combined with exercise) or no advice with either metformin (850 mg twice daily) or placebo (Hoeger et al., 2004). The greatest effect was in the combination group with respect both to reduction of weight and hyperandrogenism. Yet irrespective of treatment group the greatest improvement in rate of ovulation was achieved by those who lost weight (Hoeger et al., 2004).
There remain a number of unanswered questions concerning the use of metformin in women with PCOS, including which parameters may best predict a response and the appropriate dose for a given body mass. Metformin therapy certainly appears beneficial in certain circumstances and may alone improve menstrual cyclicity, ovulation and hyperandrogenism in some women (Costello and Eden, 2003; Lord et al., 2003). Furthermore metformin may amplify the effects of ovulation-inducing drugs (Costello and Eden, 2003; Lord et al., 2003) or androgen-lowering medication (Gambineri et al., 2004). We have found, however, that in very obese women with anovulatory PCOS, metformin, at a dose of 850 mg twice daily, had no effect on menstrual frequency, body weight or insulin sensitivity, despite a fall in total testosterone and waist circumference. Furthermore a modest reduction in weight through lifestyle modification was the most significant predictor for an improvement in menstrual cyclicity.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
We are grateful to colleagues in participating centres: Dr R.Anderson (MRC Reproductive Medicine Centre, Edinburgh), Dr B.Bentick (Consultant Gynaecologist, Royal Shrewsbury Hospital, Shrewsbury), Dr P.Hardiman (Consultant Gynaecologist, Royal Free Hospital, London), Dr N.Panay (formerly Gynaecologist, Fertility Centre, St Bartholomew’s Hospital, London), Dr H.Buckler (Consultant Physician, Hope Hospital, Salford) and Professor W.Ledger (Department of Obstetrics and Gynaecology, The Jessop Hospital for women, Sheffield). We also thank Ms M.O’Kane (Clinical Specialist Dietitian, Department of Nutrition and Dietetics, The General Infirmary at Leeds, Leeds) for her advice and efforts to encourage our subjects to lose weight. Grant: The Special Trustees of the Leeds Teaching Hospitals, NHS Trust, Leeds.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Adams J, Franks S, Polson DW, Mason HD and Jacobs HS (1985) Multifollicular ovaries: clinical and endocrine features and response to pulsatile gonadotrophin releasing hormone. Lancet 2,1375–1378.
Balen AH (1999) Pathogenesis of polycystic ovary syndrome—the enigma unravels? Lancet 354,966.[CrossRef][ISI][Medline]
Balen AH and Michelmore K (2002) What is polycystic ovary syndrome? Are national views important? Hum Reprod 17,2219–2227.
Balen AH, Conway G, Kaltsas G, Techatraisak K, Manning PJ, West C and Jacobs HS (1995) Polycystic ovary syndrome: the spectrum of the disorder in 1741 patients. Hum Reprod 10,2107–2111.
Balen AH, Laven JSE, Tan SL and Dewailly D (2003) Ultrasound assessment of the polycystic ovary: international consensus definitions. Hum Reprod Update 9,505–514.
Berek J, Adashi E and Hillard P (1996) Novak’s Gynecology. 12th edn, Williams & Wilkins, p 159.
Bonara E, Targher G, Alberiche M, Bonadonna RC, Sassian F and Zenere MB (2000) Homeostasis model assessment closely mirrors the glucose clamp technique in the assessment of insulin sensitivity: studies in subjects with various degrees of glucose tolerance and insulin sensitivity. Diabetes Care 23,57–63.[Abstract]
Cedergren MI (2004) Maternal morbid obesity and the risk of adverse pregnancy outcome. Obstet Gynecol 103,219–224.
Cibula D, Dvorakova K, Stanicka S, Sindelka G, Hill M and Fanta M (2004) Insulin sensitivity in women with polycystic ovary syndrome. J Clin Endocrinol Metab 89,2942–2945.
Clark AM, Ledger W, Galletly C, Tomlinson L, Blaney F, Wang X and Norman R (1995) Weight loss results in significant improvement in pregnancy and ovulation rates in anovulatory obese women. Hum Reprod 10,2705–2712.
Clark AM, Thornley, Tomlinson L, Galletley C and Norman R (1998) Weight loss in obese infertile women results in improvement in reproductive outcome for all forms of fertility treatment. Hum Reprod 13,1502–1505.
Conway GS (1990) Insulin resistance and PCOS. Contemp Rev Obstet Gynaecol 2,34–39.
Costello M and Eden J (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]
Crave JC, Fimbel S, Lejeune H, Cugnardy N, Dechaud H and Pugeat M (1995) Effects of diet and metformin administration on sex hormone-binding globulin, androgens and insulin in hirsute and obese women. J Clin Endocrinol Metab 80,2057–2062.[Abstract]
Crosignani PG, Colombo M, Vegetti, Somigliana E, Gessati A and Ragni G (2003) Overweight and obese anovulatory patients with polycystic ovaries: parallel improvements in anthropometric indices, ovarian physiology and fertility rate induced by diet. Hum Reprod 18,1928–1932.
Despres JP, Lemieux I and Prud’homme D (2001) Treatment of obesity: need to focus on high risk, abdominally obese patients. Br Med J 322,716–720.
Dunaif A, Xia J, Book C, Schenker E and Tang Z (1995) Excess insulin receptor serine phosphorylation in culture fibroblasts and in skeletal muscle: a potential mechanism for insulin resistance in polycystic ovary syndrome. J Clin Invest 96,801–810.
Ehrmann DA, Cavaghan MK, Imperial J, Sturis J, Rosenfield RL and Polonsky KS (1997) Effects of metformin on insulin secretion, insulin action, and ovarian steroidogenesis in women with polycystic ovary syndrome. J Clin Endocrinol Metab 82,524–530.
Fleming R, Hopkinson ZE, Wallace AM, Greer IA and Sattar N (2002) Ovarian function and metabolic factors in women with oligomenorrhoea treated with metformin in a randomized double blind placebo-controlled trial. J Clin Endocrinol Metabol 87,569–574.
Gambineri A, Pelusi C, Genghini S, Morselli-Labate AM, Cacciari M, Pagotto U and Pasquali R (2004) Effect of flutamide and metformin adminsitered alone or in combination in dieting women with polycystic ovary syndrome. Clin Endocrinol 60,241–249.[CrossRef][Medline]
Gjonnaess H (1989) The course and outcome of pregnancy after ovarian electrocautery with PCOS: the influence of body weight. Br J Obstet Gynaecol 96,714–719.[ISI][Medline]
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][ISI][Medline]
Hamilton-Fairley D, Kiddy D, Watson H, Paterson C and Franks S (1992) Association of moderate obesity with a poor pregnancy outcome in women with polycystic ovary syndrome treated with low dose gonadotrophin. Br J Obstet Gynaecol 99,128–131.[ISI][Medline]
Hanson RL, Pratley RE, Bogardus C, Narayan KM, Roumain JM and Imperatore J (2000) Evaluation of simple indices of insulin sensitivity and insulin secretion for use in epidemiological studies. Am J Epidemiol 151,190–198.
Hoeger KM, Kochman L, Wixom N, Craig K, Miller RK and Guzick DS (2004) A randomized, 48 week, placebo-controlled trial of intensive lifestyle modification and/or metformin therapy in overweight women with polycystic ovary syndrome: a pilot study. Fertil Steril 82,421–429.[CrossRef][ISI][Medline]
Huber-Buchholz MM, Carey DG and Norman RJ (1999) Restoration of reproductive potential by lifestyle modification in obese polycystic ovary syndrome: role of insulin sensitivity and luteinizing hormone. J Clin Endocrinol Metab 84,1470–1474.
Katz A, Nambi S, Mather K, Baron AD, Follmann DA and Sullivan G (2000) Quantitative insulin sensitivity check index: a simple accurate method for assessing insulin sensitivity in humans. J Clin Endocrinol Metab 85,2402–2410.
Kauffman RP and Castracane D (2003) Controlling PCOS. Part 1: Assessing insulin sensitivity. Contemp Obstet Gynaecol 48,30–48.
Kiddy DS, Sharp PS, White DM, Scanlon MF, Mason HD, Bray CS and Franks S (1990) Differences in clinical and endocrine features between obese and non-obese subjects with polycystic ovary syndrome: an analysis of 263 consecutive cases. Clin Endocrinol 32,213–220.[Medline]
Kiddy DS, Hamilton-Fairley D, Bush A, Anyaoku V, Reed MJ and Franks S (1992) Improvement in endocrine and ovarian function during dietary treatment of obese women with polycystic ovary syndrome. Clin Endocrinol 36,105–111.[Medline]
Kirchengast S and Huber J (2004) Body composition characteristics and fat distribution patterns in young infertile women. Fertil Steril 81,539–544.[CrossRef][ISI][Medline]
Kolstad HA, Bonde JP, Hjællund NH, Jensen TK, Henrikesen TB and Olsen J (1999) Menstrual cycle pattern and fertility: a prospective follow-up study of pregnancy and early embryonal loss in 295 couples who were planning their first pregnancy. Fertil Steril 71,490–496.[CrossRef][ISI][Medline]
la Marca A, Morgante G, Palumbo M, Cianci A, Petraglia F and De Leo V (2002) Insulin-lowering treatment reduces aromatase activity in response to follicle-stimulating hormone in women with polycystic ovary syndrome. Fertil Steril 78,1234–1239.[CrossRef][ISI][Medline]
Linder A (1970) Planen und Auswerten Versuchen. Birkhauser, Basle.
Linné Y (2004) Effects of obesity on women’s reproduction and complications during pregnancy. Obes Rev 5,137–143.[CrossRef][Medline]
Lord J and Wille T (2002) Polycystic ovary syndrome and fat distribution: the central issue? Hum Fertil 5,67–71.
Lord JM, Flight IH and Norman RJ (2003) Insulin-sensitising drugs (metformin, troglitazone, rosiglitazone, pioglitazone, d-chiro-inositol) for polycystic ovary syndrome. The Cochrane Database of Systematic Reviews 2003, Issue 2. Art. No.: CD003053. DOI: 10.1002/14651858.CD003053.
Machin D and Campbell M (1987) Statistical Tables for the Design of Clinical Trials. Blackwell, Oxford.
Maciel GAR, Junior JMS, Motta ELA, Haida MA, Lima GR and Baracat EC (2004) Nonobese women with polycystic ovary syndrome respond better than obese women to treatment with metformin. Fertil Steril 81,355–360.[CrossRef][ISI][Medline]
Mansfield R, Galea R, Brincat M, Hole D and Mason H (2003) Metformin has direct effects on human ovarian steroidogenesis. Fertil Steril 79,956–962.[CrossRef][ISI][Medline]
Mathur KJ, Hunt AE, Sternberg HO, Paradisi G, Hook G, Katz A et al (2001) Repeatability characteristics of simple indices of insulin resistance: implications for research applications. J Clin Endocrinol Metab 86,5457–5464.
Matthaei S, Stumvoll M, Kellerer M and Haring HU (2000) Pathophysiology and pharmacological treatment of insulin resistance. Endocr Rev 21,585–618.
Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF and Turner RC (1985) Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28,412–419.[CrossRef][ISI][Medline]
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 85,139–146.
Morales AJ, Laughlin GA, Butzow T, Maheshwari H, Baumann G and Yen SSC (1996) Insulin, somatotropic and luteinizing hormone axes in lean and obese women with polycystic ovary syndrome: common and distinct features. J Clin Endocrinol Metab 81,3854–2864.
Moran L and Norman RJ (2004) Understanding and managing disturbances in insulin metabolism and body weight in women with polycystic ovary syndrome. Best Pract Res Clin Obstet Gynaecol 18,719–736.[CrossRef][Medline]
Moran LJ, Noakes M, Clifton PM, Tomlinson L and Norman RJ (2003) Dietary composition in restoring reproductive and metabolic physiology in overweight women with polycystic ovary syndrome. J Clin Endocrinol Metab 88,812–819.
Morin-Papunen LC, Koivunen RM, Ruokonen A and Martikainen HK (1998) Metformin therapy improves the menstrual pattern with minimal endocrine and metabolic effects in women with polycystic ovary syndrome. Fertil Steril 69,691–696.[CrossRef][ISI][Medline]
Munster K, Schmidt L and Helm P (1993) Length and variation in the menstrual cycle—a cross-sectional study from a Danish county. Br J Obstet Gynaecol 99,422–429.
Norman RJ, Davies MJ, Lord J and Moran LJ (2002) The role of lifestyle modification in polycystic ovary syndrome. Trends Endocrinol Metab 13,251–257.[CrossRef][ISI][Medline]
Norman RJ, Noakes M, Wu R, Davies MJ, Moran L and Wang JX (2004) Improving reproductive performance in overweight/obese women with effective weight management. Hum Reprod Update 10,67–280.
Pasquali R, Antenucci D, Caimirri F, Venturoli S, Paradis R and Fabbri R (1989) Clinical and hormonal characteristics of obese amenorrheic hyperandrogenic women before and after weight loss. J Clin Endocrinol Metab 68,173–179.[Abstract]
Pasquali R, Gambineri A, Biscotti D, Vincennati V, Gagliardi L and Colitta D (2000) Effect of long-term treatment with metformin added to hypocaloric diet on body composition, fat distribution and androgen and insulin levels in abdominally obese women with and without polycystic ovary syndrome. J Clin Endocrinol Metab 85,2767–2774.
Pasquali R, Pelusi C, Genghini S, Cacciari M and Gambineri A (2003) Obesity and reproductive disorders in women. Hum Reprod Update 9,359–372.
Pirwany IR, Yates RW, Cameron IT and Fleming R (1999) Effects of the insulin sensitizing drug metformin on ovarian function, follicular growth and ovulation rate in obese women with oligomenorrhoea. Hum Reprod 14,2963–2968.
Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group (2004) Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome. Hum Reprod 19,41–47 and Fertil Steril 81,19–25.
Sebire NJ, Jolly M, Harris JP, Wadsworth J, Joffe M, Beard RW, Regan L and Robinson S (2001) Maternal obesity and pregnancy outcome: a study of 287,213 pregnancies in London. Int J Obes Relat Metab Disord 25,1175–1182.[CrossRef][ISI][Medline]
Stamets K, Taylor DS, Kunselman A, Demers LM, Pelkman CL and Legro RS (2004) A randomized trial of the effects of two types of short-term hypocaloric diets on weight loss in women with polycystic ovary syndrome. Fertil Steril 81,630–637.[CrossRef][ISI][Medline]
Tsilchorozidou T, Overton C and Conway GS (2004) The pathophysiology of polycystic ovary syndrome. Clin Endocrinol 60,1–17.[CrossRef][Medline]
Velazquez EM, Mendoza S, Hamer T, Sosa F and Glueck CJ (1994) Metformin therapy in polycystic ovary syndrome reduces hyperinsulinaemia, insulin resistance, hyperandrogenaemia and systolic blood pressure, while facilitating normal menses and pregnancy. Metabolism 43,647–654.[CrossRef][ISI][Medline]
Velazquez EM, Acosta A and Mendoza SG (1997a) Menstrual cyclicity after metformin therapy in PCOS. Obstet Gynecol 90,392–395.[Abstract]
Velazquez EM, Mendoza SG, Wang P and Glueck CJ (1997b) Metformin therapy is associated with a decresae in plasminogen activator inhibitor-1, lipoprotein(a) and immunoreactive insulin levels in patients with PCOS. J Clin Endocrinol Metab 82,524–530.
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.
Zaadstra BM, Seidell JC, Van Noord PA, te Velde ER, Habbema JD and Vrieswijk B (1993) Fat and female fecundity: prospective study of effect of body fat distribution on conception rates. Br Med J 306,484–487.
Submitted on April 25, 2005; resubmitted on June 29, 2005; accepted on June 30, 2005.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. L. Rofey, E. M. Szigethy, R. B. Noll, R. E. Dahl, E. lobst, and S. A. Arslanian Cognitive-Behavioral Therapy for Physical and Emotional Disturbances in Adolescents with Polycystic Ovary Syndrome: A Pilot Study J. Pediatr. Psychol., June 21, 2008; (2008) jsn057v2. [Abstract] [Full Text] [PDF]
|
|