Effects of metformin on polycystic ovary syndrome: a randomized, double-blind, placebo-controlled study

Background and objectives: Metformin improves manifestations of polycystic ovary syndrome (PCOS) by reducing insulin resistance. The objective of this study was to determine how metformin, in combination with lifestyle changes, affects the clinical manifestations of PCOS. Materials and Methods: Patients with PCOS attending the outpatient of a tertiary care hospital were enrolled in the study. Revised Rotterdam Consensus 2003 criteria were used to diagnose cases of PCOS. Clinical information, anthropometric measurement, serum progesterone and polycystic ovarian morphology (PCOM) of each subject were recorded in a prescribed data sheet at baseline and after a period of nine months. Randomized placebo controlled double blind design was used to assign participants in respective groups. Participants were randomly assigned to receive 9 month course of either metformin (1500 mg/day) or placebo. Both groups were advised regarding schedule of lifestyle modification. Outcome variables were clinical manifestations related to metabolic, reproductive and androgenic status of PCOS. Results: Out of 80 enrolled PCOS cases, 49 completed the study (metformin=26, placebo=23). The mean age of the study participants of metformin and placebo groups was 23.52±5.18 and 22.09±3.58 years respectively (p=0.262). Menstrual cycle significantly improved in both the study groups (before vs. after metformin: 19.2% vs. 76.9%, p=0.003; placebo: 19.2% vs. 47.8%, p=0.02) after 9 months, but compared to placebo group no such significant (p=0.12) improvement occurred in metformin group. Severity of hirsutism, presence of acne, serum progesterone level and ovulatory status improved significantly in both groups after completion of the study. Except acanthosis nigricans, other metabolic manifestations did not significantly improve in metformin compared to placebo group after the intervention. While comparing the percentage changes, body mass index (BMI) and waist circumference (WC) reduced significantly in metformin than placebo group (BMI in kg/m 2 metformin vs. placebo: -3.63±8.22 vs. +1.42±6.67, p= 0.024; WC in cm 2.81±7.74 vs. +1.68±7.89, p= 0.05). No significant adverse event was observed in metformin group. Conclusion: Metformin, in conjunction with lifestyle modifications, has favorable impacts on clinical manifestations of PCOS. IMC J Med Sci 2021; 15(2): 001

PCOS has been associated with the risk of type 2 diabetes mellitus (T2DM) and metabolic syndrome [12]. Insulin resistance and obesity are the common features seen in PCOS women. About 60-70% and up to 88% women with PCOS are reported to have insulin resistance and obesity respectively [13]. Recent evidences have shown an important implication of insulin resistance in the pathogenesis of PCOS irrespective of obesity or over weight [14,15]. So, the presence of insulin resistance has prompted widespread use of insulin sensitizing agents in the treatment of PCOS [16]. Among the insulin sensitizers, metformin is the most commonly used agent for the treatment of PCOS [17]. In PCOS patients, metformin has been reported to act by restoring ovulation, and by reducing the weight, circulating androgen levels, risk of miscarriage and gestational diabetes mellitus [13]. Metformin is more effective in reducing hirsutism also [18]. Pioglitazone, another widely used insulin sensitizer, is more effective in reducing menstrual irregularity and ovulation (odds ratio= 2.31) than metformin, but associated with weight gain. Apart from the use of insulin sensitizing agents, lifestyle modifications also have therapeutic value in the management of clinical manifestations of PCOS. It helps especially in reduction of weight, waist circumference and hirsutism [19]. Combination effects of lifestyle modification and insulin sensitizer have been reported to be better than each modality of treatment alone especially in reducing body mass index (BMI) and improving menstrual irregularity [20]. In view of the above, the present study was undertaken to determine whether metformin or placebo, in conjunction with lifestyle changes, could affect the clinical manifestations of patients with PCOS.

Materials and methods
The research protocol was approved by the Institutional Review Board (IRB) of BSMMU (No.BSMMU/2014/05). Informed written consent was obtained from all participants. At the end of the study period, PCOS patients in placebo group were offered treatment with metformin based on the study findings. ] formula based on the mean±SD changes of BMI from a similarly designed study [22].
Intervention: After confirmation of diagnosis, patients were randomly allocated to either metformin or placebo group by lottery method in a double blind process. Both groups were provided advice regarding schedule of lifestyle modification. Each patient was advised regarding lifestyle modifications that included a weight-based diet [calorie calculated (weight maintenance Kcal ± 500 Kcal) and diet chart provided], brisk walking for at least 30 minutes at least five days a week and behavioral modification as appropriate [23]. Metformin was prescribed as 1500 mg/day in three divided doses for 9 months. Each tablet of metformin contained metformin HCl BP 500 mg, along with inactive ingredients of povidone, microcrystalline cellulose, croscarmellose sodium and magnesium stearate as excipient. Placebo tablet contained all the ingredients of excipient except metformin. Both metformin and placebo tablets were coded and dispensed in identical blister package.

Follow-up schedule:
Each patient was scheduled for five visits within a period of nine months whereupon a window period of 14 days was permitted for the 2 nd to 5 th visits. In each follow up visit, patients were asked about the practice of lifestyle modification schedules and intake of investigational product (IP). Patients who did not follow all the schedules of lifestyle modifications at least five days a week were considered as noncompliant. Similarly, they were also asked to bring the empty blisters of the IP. The blisters were counted and the participants who missed to take ≥5% of the IP were also labeled as noncompliant. Noncompliant participants were excluded from final analysis.

Outcome variables:
The main outcome variables were metabolic (BMI, WC, blood pressure), reproductive (menstrual cycle, ovulation by progesterone level and polycystic ovarian morphology) and androgenic (modified Ferriman-Gallwey score and acne and acanthosis nigricans) manifestations of PCOS. Weight (kilogram) and height (meter) were measured by calibrated bathroom scale and mounted measuring tape respectively to calculate BMI (kg/m 2 ). Waist circumference (in centimeter) was measured by measuring tape at the level of umbilicus. Asian criteria was used to categorize BMI as under nutrition, optimal, overweight and obese by the cut-offs values of 18.5, 23 and 25 kg/m 2 respectively, and WC by the cut-off value of 80 cm [24]. Blood pressure was measured by calibrated sphygmomanometer (mm-Hg) and was categorized by metabolic syndrome criteria (cut-off of 130/85 mm-Hg) [25]. Hirsutism was measured by using modified Ferriman-Gallway (mFG) score and classified into mild, moderate and severe with the total score of 8 -15, 16 -25 and ≥26 respectively [26]. Acne was observed over face. Acanthosis nigricans was checked on neck, axilla and groin. Missing of at least three menstrual periods in a row was labeled as amenorrhea and without menstruation regularly for more than 35 days termed as oligomenorrhea. Menstruation occurring for consecutive two months was considered as regular menstruation.
Investigations: Blood samples from each participant were collected aseptically for biochemical investigations at the time of enrollment and after nine months. Fasting blood glucose (FBG) was measured by glucose oxidase method and serum luteinizing hormone (LH), follicle stimulating hormone (FSH), total testosterone (TT), and progesterone were measured by chemiluminescent microparticle immunoassay at diagnosis during follicular phase of menstrual cycle. Serum progesterone ≥5 ng/ml was considered as indication of ovulation [27]. Ultrasonogram (TOSHIBA Aplio 500 imaging machine) was done by trans-abdominal or trans-vaginal route depending on the marital status of the patient during follicular phase of menstrual cycle by skilled sonologists.
Data recording: A standardized questionnaire was used to obtain information regarding the age, personal and family history, obstetric, menstrual and gynecological history, medical diseases, and medications. Clinical information, anthropometric measurement, results of biochemical tests and polycystic ovarian morphology (PCOM) of each subject were recorded in a prescribed data sheet at baseline and after nine months during the last follow up visit (4 th ). Compliance and side-effects were recorded in all the follow up visits.
Data analysis: All data were processed by the SPSS program (version: 22.0). Values were expressed as frequencies (%) or meanSD or median (interquartile range, IQR). Nominal variables namely occupation, personal and family history, acne, acanthosis nigricans, WC, BP and ovulation categories, inference of ultrasonography findings and side-effects at each visit were analyzed by using Pearson's Chi square/ Fisher's exact test between groups and McNemer test (two related samples) within each group (intervention: before vs. after). Ordinal variables (menstrual cycle, BMI category and severity of hirsutism) were tested by Chi-square test/Fisher's exact test between groups and Wilcoxon signed ranks test within group. Normally distributed quantitative variables (age, BMI, WC, BP, FBG, biochemical results) were analyzed by independent student's t test between groups and paired-t test within group. Skewed variables (mFG score, progesterone, TSH, prolactin, LH/FSH) were analyzed by Mann-Whitney-U test between groups and Wilcoxon rank sum test within each group. A two-tailed p<0.05 was taken as statistically significant.

Results
Initially, a total 80 women with PCOS were enrolled in the study. Out of 80, 49 PCOS patients completed the study. After unblinding, it was observed that 26 cases were in metformin group and 23 were in placebo group. Out of the rest 31 cases, 4 participants became pregnant before reaching the end point (Metformin = 1, placebo = 3), and 27 were dropped out due to incomplete visit (metformin= 9, placebo= 9) or noncompliant to lifestyle modification or intake of IP (metformin = 4, placebo = 5) (Figure-1).
The mean age of the total study participants of metformin and placebo groups was 23.52±5.18 and 22.09±3.58 years respectively (p=0.262). Other baseline characteristics of the two groups were similar with respect to occupation, personal and family history and biochemical parameters ( Table 1). Table-2 shows the status of metabolic manifestations in PCOS patients before and 9 months after intervention. Compared to placebo group, the metabolic manifestations, except acanthosis nigricans, did not significantly improve in metformin group 9 months after the intervention. Compared to metformin group, the rate of acanthosis nigricans was significantly higher in placebo group at baseline (metformin vs. placebo: 19.2% vs. 52.2%, p=0.02) and remained significantly higher after the end of intervention also (7.7% vs. 39.1%, p=0.015). BMI (25.17±5.62 vs.        .001] than placebo group. Other adverse effects were similar in rates in both groups. Frequency of loose motion and nausea improved in metformin group and abdominal pain reduced in placebo group with progression of time (Table-4).

Discussion
This randomized placebo controlled double blind study showed the beneficial effects of addition of metformin with lifestyle modifications on reduction of BMI and WC without causing significant sideeffects in patients with PCOS. We observed that metformin with lifestyle modification led to the reduction of BMI. Most of the studies found similar results and few studies also found that metformin reduced insulin resistance even without reducing BMI, in PCOS [22]. Pouliot et al. showed that higher WC is predictive of abnormal endocrine and metabolic function and is associated with increasing risk of cardiovascular disease [28]. In our study, the mean reduction of WC in metformin group was 2.81 cm. Other investigators have also reported reduction of WC, though some reported no changes in WC after treatment with metformin [17,29,30]. We found that metformin, in combination with lifestyle changes, significantly reduced BMI and WC especially in obese patients. A recently published systematic review and metaanalysis of randomized controlled trials (RCTs) showed that metformin significantly reduced WC in PCOS patients [31].
In the present study, there was significant improvement of menstrual irregularity in both metformin and placebo groups. Therefore in the present study, it was not clearly possible to discriminate the impact of metformin on this issue as lifestyle modification was practiced by the  [29,30,[32][33][34]. However, improvement observed in the placebo group indicated the role of lifestyle modification even without significant change in BMI. Some studies found higher regularization of menstrual cycle by modification of lifestyle than metformin alone. However, the combination therapy seemed to be better than the lifestyle therapy alone.
In the present study hirsutism improved in both metformin and placebo groups. However, we observed significantly greater decrease of mFG score in metformin compared to placebo group after 9 months of study period. The results indicated that metformin and lifestyle modification together had better effect on hirsutism rather than either one alone. Contrary to our findings, some studies reported lack of beneficial effect of metformin on hirsutism in PCOS [33,34]. However, those studies observed the effect of metformin on women with high baseline mFG score and for a period of only six months. The effect of metformin on the hair follicles to reduce the state of hirsutism might not be apparent if the treatment duration is below six months as the hair turnover cycle is more than three months. Therefore, it may not be possible to make any inference on the action of metformin in reducing hirsutism unless treatment for adequate period is provided.
As in our cases, improvement of acne in metformin group is also supported by observations of others [32,35]. There was no significant change of polycystic appearance of ovaries in either group after nine months of study period. However, ultrasonogram of ovaries lacked description of status of ovarian follicles.
Four women conceived during the study, three of which were in the placebo group and one in the metformin group. Assessment of impact of metformin on fertility probably requires a longer duration of follow-up. One study found that after six months of metformin therapy, menstrual regularity was achieved in 86% women and 50% were ovulatory at six months [36]. We also found significantly higher rate of ovulation in the metformin group. Costello and Eden, also found similar improvement in obese PCOS patients [37].
There were no serious adverse events in either the metformin or placebo group. However, metformin was associated with more minor gastrointestinal events. But, the adverse events gradually improved over time.
The main strength of this study was its randomized double blind placebo controlled design and its long duration. The study findings also illustrated the importance of life style modification in management of clinical manifestations of PCOS. In conclusion, the present study showed that metformin with lifestyle modifications had considerable favorable impacts on clinical manifestations, particularly in terms of androgenic activity, weight loss, WC, BMI, and menstrual irregularity in PCOS patients. Effect of metformin therapy on fertility of PCOS cases needs longer duration of follow-up.