The use of metformin in women with polycystic ovary syndrome: an updated review

Abstract

Purposes

Polycystic ovary syndrome (PCOS) is a major cause of female infertility, being present in up to 20% of women of childbearing age. Insulin resistance (IR) plays an important role in the pathophysiology of PCOS; therefore, its treatment may benefit women with the syndrome. The main drug used for IR management is metformin (MT). We aim to review the literature on the use of metformin in women with PCOS.

Methods

Using the terms “metformin” and “polycystic ovary syndrome,” we conducted a search the PubMed, EMBASE, and Google Scholar databases. The research was restricted to articles published in English. Initially, only published meta-analyses were included, in the absence of meta-analyzes, RCT and well-designed prospective studies were used.

Results

Metformin increases success rates and decreases complication rates when used as an adjunctive medication for ovulation induction during low complexity assisted reproduction treatments and during ovarian stimulation for in vitro fertilization in women with PCOS. Evidence about the effect of metformin on fetal and obstetric complication rates is conflicting. Metformin is associated with high incidence of gastrointestinal symptoms; however, serious adverse effects are rare and there is no evidence of teratogenicity.

Conclusion

For women with PCOS, metformin is a good adjunctive medication for ovulation induction/stimulation for high and low complexity assisted reproduction therapies. The adverse effects are mostly mild, and there is no risk of teratogenicity, but the risk of long-term complications for the offspring is not yet defined. High heterogeneity of the studies limits extrapolation of findings, and further research is needed to determine which women will benefit most from the medication.

Introduction

Polycystic ovary syndrome (PCOS) is a reproductive and metabolic disorder characterized by the presence of menstrual irregularity and hyperandrogenism. The prevalence of PCOS ranges from 2 to 26% of the women at reproductive age [1–3]; therefore, it is considered the most common endocrinopathy in women and a frequent cause of female infertility [2, 4]. The clinical presentation of PCOS is varied, including several degrees of menstrual cycle changes (amenorrhea, oligomenorrhea), which can cause infertility secondary to anovulation, hirsutism (increased hair growth in male regions of the body), acne, obesity, and metabolic disorders (insulin resistance, type 2 diabetes mellitus) [5]. Oligomenorrhea or anovulation are the most frequent findings of PCOS, affecting up to 74% of the women [5]. Despite the recommendation to classify PCOS patients according to phenotypes, these women often present with mild signs and symptoms only [6].

Hyperandrogenism, the main metabolic disorder of PCOS, is a consequence of excessive production of dehydroepiandrosterone and androstenedione by the teca cells and the adrenal glands [7]. In addition, women with PCOS often present with insulin resistance (IR), characterized by high insulin basal secretion and decreased insulin response to glucose overload. Insulin excess acts synergistically with luteinizing hormone, stimulating androgen production by teca cells, further aggravating the hyperandrogenism [3]. Moreover, the adipose tissue of women with PCOS contribute to IR due to adipocytokine dysfunction, decreased secretion of adiponectin, dysregulation of free fatty acid metabolism, and epigenetic abnormalities that affect the function of the glucose transporter 4, an important regulator of insulin sensibility [3, 8].

Due to the central role of IR in the pathophysiology of PCOS, there is biological plausibility for IR treatment in women with PCOS with the objective of reducing the metabolic disorders and promoting the return of normal ovulatory function. Among the treatment options for IR, MT has been the most studied. The mechanisms of action of MT are not completely understood, but include reduction of the hepatic gluconeogenesis through its direct action on the liver cells [9]. Moreover, in vitro studies using human and other mammalian germ cells have shown that MT is able to regulate androgens and estradiol production as well as follicular recruitment [10, 11]. Therefore, MT has been used in women with PCOS for infertility treatment and prevention of obstetric and fetal complications. In this review, we aimed to summarize the current literature about MT use in women with PCOS.

Methods

We conducted a narrative review about MT use in women with PCOS. We included studies comparing the effect of MT or combined treatments including MT versus placebo or standard treatment in the following scenarios: ovulation induction (OI), ovarian hyperstimulation during in vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI), prevention and treatment of gestational diabetes mellitus (GDM), and prevention of fetal complications. PCOS definition was based on the Rotterdam ESHRE/ASRM consensus criteria, NIH criteria, or AES criteria. We focused on the most recent meta‑analyses (MA) evaluating these topics, and when MAs regarding a specific topic were lacking, we included randomized controlled trials (RCT) or high-quality prospective studies.

We conducted an electronic search using PubMed/MEDLINE, EMBASE, and Cochrane’s centra database until December 2021. The search combined terms and descriptors were “metformin” AND “polycistic ovary syndrome.” We also searched among the references of the identified articles. There were no limits placed on the year of publication, but we restricted the search to articles published in English. The following outcomes were assessed when applicable: ovulation rate, clinical pregnancy rate (CPR), live birth rate (LBR), ovarian hyperstimulation syndrome (OHSS) rate, and adverse effects (AE) rate. This study was exempted from Institutional Review Board approval as it did not involve any human intervention.

Results

Metformin use in ovulation induction of infertile women with PCOS

Regarding the use of MT for OI during timed sexual intercourse, several MAs have evaluated both the isolated use of MT and the combination of MT with other drugs. The most recent Cochrane MA selected only RCTs that defined PCOS based on the Rotterdam Consensus criteria and reported reproductive outcomes [12]. This MA demonstrated that MT was more effective in inducing ovulation (13 RCTs, 684 women; OR 2.64, 95% CI 1.85 to 3.75; I² = 61%; low‐quality evidence), with higher CPR (11 RCTs, 1213 women; OR 1.98, 95% CI 1.47 to 2.65; I² = 30%; moderate‐quality evidence) and LBR (4 RCTs, 435 women; OR 1.59, 95% CI 1.00 to 2.51; I² = 0%; low‐quality evidence) when compared to placebo. The same MA reported that the combination of MT and clomiphene citrate (CC) was superior to the use of CC alone, with higher rates of ovulation (21 RCTs, 1568 women; OR 1.65, 95% CI 1.35 to 2.03; I² = 63%; low‐quality evidence) and CPR (19 RCTs, 1790 women; OR 1.62, 95% CI 1.32 to 1.99; I² = 31%; moderate‐quality evidence), but without statistical difference regarding LBR (10 RCTs, 1219 women; OR 1.27, 95% CI 0.98 to 1.65; I² = 28%; low‐quality evidence). However, when MT was compared head-to-head to CC, the findings were inconclusive and inconsistent due to the high heterogeneity of the included RCTs.

Another MA from a different group was performed using individual participant data from RCTs [13]. In this MA, when the combination of MT plus CC was compared to CC alone, the authors found increased CPR (5 RCTs, 907 women; RR 1.08, 95% CI 0.87 to 1.35, I² = 5.6%, low-quality evidence), reduced time-to-pregnancy (7 RCTs, 898 women; HR 1.25,95% CI 1.00 to 1.57, I² = 0, low-quality evidence), and increased LBR (5 RCTs, RR 1.08, 95% CI 0.87 to 1.35, I² = 5.6%; low-quality evidence). In addition, there was a positive interaction between baseline insulin levels and the treatment effects on LBR, suggesting that women with higher baseline serum insulin levels have larger treatment effects with the combined therapy. Moreover, when MT was compared to CC in a MA including only non-obese (body mass index < 32 kg/m²) women, there were no differences regarding ovulation rate per cycle (2 RCT, 909 cycles; RR 0,79, 95% CI 0.54 to 1.17, I² = 78%), CPR ( 4 RCT, 465 women; RR 0,98, 95% CI 0.49 to 1.96, I² = 80%), and LBR ( 3 RCT, 285 women; RR 0,84, 95% CI 0.22 to 6.26, I² = 90%)[14]. These findings suggest that MT is more beneficial in women with PCOS who have IR. In the studies included in these MAs, MT was used at doses ranging from 850 to 2000 mg/day continuously throughout the study.

Another scenario assessed was the comparison between the association of MT and CC versus other drugs in women with PCOS refractory to CC, defined as persistent anovulation following treatment with 150 mg of CC daily for three consecutive attempts. The most comprehensive MA on this subject included only RCTs that defined PCOS based on the Rotterdan consensus, on the National Institute of Health criteria, or on the Androgen Excess Criteria [15]. The study reported that ovulation rate (3 RCTs, 323 women; OR 0.25, 95% CI 0.15 to 0.41, I² = 85%) and CPR (3 RCTs, 323 women; OR 0.45 95% CI 0.27 to 0.75, I² = 0%) were lower when the MT plus CC was compared to the use of gonadotropins. However, when the combined therapy was compared to the use of aromatase inhibitors, there were no differences in the outcomes (ovulation rate: 3 RCTs, 409 women, OR 0.88, 95% CI 0.58 to 1.34, I² = 3%; CPR: 2 RCTs, 309 women; OR 0.85, 95% CI 0.53 to 1.37). LBR was also lower with the combination therapy when compared to gonadotropins (2 RCTs, 170 women; OR 0.33 95%, CI 0.13 to 0.85, I² = 0%), and when compared to aromatase inhibitors in one small trial (OR = 0.21; 95%CI 0.05–0.87; p = 0.03). The authors did not comment on the dose regimen used in the studies.

Furthermore, MT has been used in combination with gonadotropins in CC-refractory PCOS. The first MA comparing the combination of MT plus gonadotropins versus gonadotropins alone included 7 studies and found higher CPR (7 RCTs, 942 women; OR = 2.25, 95% CI 1.50 to 3.88, I² = 0%) and higher LBR (2 RCTs, 661 women; OR = 1.94, 95% CI 1.10 to 3.44, I² = 30%) with the combination therapy [16]. A recent Cochrane MA with more strict inclusion criteria also reported higher CPR (5 RCTs, 264 women; OR = 2.51, 95% CI 1.46 to 4.31, I² = 0%; low-quality evidence) and higher LBR (2 RCTs, 180 women; OR = 2.31, 95% CI 1.23 to 4.34, I² = 0%; low-quality evidence). In these studies, MT was used at a dose of 1500 to 1700 mg/day, started 4 to 12 weeks before ovulation induction and continued until ovulation triggering or pregnancy test in most studies.

Metformin has also been compared to lifestyles changes (LSC). The most recent MA regarding this topic included 13 prospective studies and assessed menstrual cycle frequency, as a surrogate for ovulation induction, and pregnancy rate [17]. The most frequent LSC reported were hypocaloric diet and exercise during 3 to 12 months, and the MT dose regimen varied from 1700 to 2000 mg/day. When LSC was compared to MT alone, there were no differences in menstrual frequency (4 studies, OR 1.62, 95% CI 0.98 to 2.69, I² = 0) and CPR (2 studies, OR 1.44, 95% CI 0.72 to 2.85, I² = 0). Likewise, when the combination of LSC and MT was compared to LSC alone, the authors did not find differences in menstrual frequency (4 studies, OR 1.20, 95% CI 0.71 to 2.02, I² = 54), whereas pregnancy rate was not analyzed. A previous MA, including RCTs comparing LSC and MT, found that the LSC plus MT group had greater number of menstrual cycles over 6 months compared with to LSC alone (3 studies, MD 1.06, 95% CI 0.30 to 1.82, I² = 0) [18]. In this analysis, LSC were diet alone in 2 studies and diet plus exercise in 1 study, while the MT dose regimen was 850 mg twice a day in all studies. Both MAs included studies with moderate to high risk of bias; therefore, their results should be taken with caution.

Metformin use in ovarian hyperstimulation for in vitro fertilization

Metformin is also used as an adjunct drug in ovarian hyperstimulation protocols for IVF/ICSI in women with PCOS. In this context, a recent Cochrane MA assessed the results of 13 RCTs, including 10 studies with long GnRH-agonist protocols, and 3 studies using short GnRH-antagonist protocols [19]. When MT was used in addition to long GnRH-agonist protocols, there was higher CPR (10 RCTs, 915 women; OR = 1.32, 95% CI 1.08–1.63, I² = 13%; low-quality evidence) in comparison with placebo, but the LBR did not increase (6 RCTs, 651 women; OR = 1.30, 95% CI 0.94–1.79, I² = 47%; low-quality evidence). In addition, MT use decreased the chance of ovarian hyperstimulation syndrome (OHSS) (9 RCTs, 898 women; OR = 0.40, 95% CI 0.26–0.60, I² = 13%; low-quality evidence). In the analysis of studies using short GnRH-antagonist protocols, CPR (2 RCTs, 177 women; OR = 1.38, 95% CI 0.21–9.14; low-quality evidence), LBR (1 RCTs, women; OR = 0.48, 95% CI 0.29–0.79; low-quality evidence), and OHSS rate (2 RCTs, 193 women; OR = 0.97 95% CI 0.32–2.98, I² = 26%; low-quality evidence) did not improve with MT use; however, the low number of studies prevented any definitive conclusion.

Another recent MA from a different group included 12 RCTs and performed a post hoc subgroup analysis based on body mass index (BMI) [20]. Overall, MT used did not increase the CPR (11 RCTs, 1015 women; OR = 1.24, 95% CI 0.82–1.86; I² = 55%) or the LBR (8 RCTs, 811 women; OR = 1.23 95% CI 0.74–2.04, I² = 62%). Like the previous MA, the authors found reduced risk of OHSS with MT (11 RCTs, 947 women; OR = 0.43 95% CI 0.24–0.78; I² = 38%). In the subanalysis by BMI, the subgroup with BMI ≥ 26 kg/m² showed increased CPR with MT use (6 RCTs, 482 women; OR = 1.71 95% CI 1.12–2.60; I² = 10%), but no differences in LBR in both groups. This MA included only 2 RCT that used short GnRH-antagonist protocols and did not perform subgroup analysis based on the type of ovarian hyperstimulation protocol. When used during IVF/ICSI, MT dose ranged from 1000 to 2550 mg/day. Most studies started MT before ovarian stimulation (16 week before earliest) and continued at least until the day of the trigger.

Metformin use for prevention of obstetric and fetal complications in women with PCOS

In addition to its role in the treatment of PCOS-related female infertility, MT has also been used during pregnancy with the aim of reducing the risk of obstetric and fetal complications resulting from IR during pregnancy. The two most recent MAs on this topic demonstrated that the use of MT during pregnancy was able to decrease the risk of early pregnancy loss in 80%, and the risk of premature birth in 70%, with very similar effect sizes in both MAs [21, 22]. However, most studies included in these two analyses were non-randomized, thus the quality of evidence is low.

Furthermore, the benefit of MT in lowering the risk of pre-eclampsia, GDM, and gestational hypertension was unclear. In the analyses including all the studies, both MAs reported a decrease of approximately 70% in the risk of GDM, but when only RCTs were evaluated, there was no statistical difference between MT and placebo. Regarding the risk of pregnancy induced hypertension/pre-eclampsia, the evidence is even more conflicting. One MA showed a benefit with MT only when non-randomized studies were included [21], whereas the second MA reported a benefit with MT; however, it did not describe the studies included in this analysis [22]. Likewise, other two MAs evaluating the role of MT in decreasing the chance of GDM also reported conflicting findings and no benefit when only RCTs were analyzed [23, 24]. The included studies in these analyses compared women that continued MT use thought pregnancy with women that either never used MT or stopped MT use during pregnancy, usually in the first trimester. Metformin was used throughout pregnancy at doses ranging from 1000 to 2000 mg/day. In addition, there was considerable variation in the definition of the primary and secondary outcomes in the included studies. Most studies were powered using the incidence of GDM as the primary outcome, whereas others used pregnancy loss. These aspects introduced high heterogeneity to the MAs, and may hinder the generalization of their results.

Moreover, a long-term study demonstrated that the use of MT during pregnancy only was not able to improve the BMI and metabolic profile of women with PCOS in the postpartum period [25]. Similarly, two MAs that assessed the use of MT in nonpregnant women with PCOS showed that its use improved clinical parameters such as BMI, waist-hip ratio, and diastolic blood pressure, but without impact on metabolic parameters and endocrine disorders [26, 27].

Adverse effects of metformin use in women with PCOS

Despite the benefits of MT for various aspects in women with PCOS, its adverse effects (AE) often lead to low compliance or interruption of the treatment. Nausea, diarrhea, and vomiting are the main AE, occurring in up to 30% of cases, and caused by decreased glucose absorption in the gastrointestinal tract [16]. Few studies have reported the incidence AE in patients with PCOS, with only 3 MAs reporting quantitative data [12, 19, 28]. When MT was used alone, the incidence of AEs ranged from 6 to 40%, with higher incidence in non-obese women [12]. The same trend appeared when the incidence of AE was compared between the combination of MT and CC versus CC alone, with higher incidence in the MT plus CC group, especially in the non-obese group [12]. In the setting of ovarian hyperstimulation for IVF/ICSI, the risk of AE was 3 × higher when MT was compared to placebo, with no difference between short GnRH-antagonist or long GnRH-agonist protocols [19]. More serious AE such as lactic acidosis and liver failure have not been reported in women with PCOS [29]. As strategies for decreasing AEs, prolonged-release presentations should be preferred, the medication should be taken on an empty stomach, and the dose should be scaled up to the maximum dose, usually 2000 mg/day, over a period of 4 to 6 weeks [29].

The impact of metformin use during pregnancy on offspring health

Another concern about MT use during pregnancy is the possible negative effects in the offspring. Although the use of MT is not associated with increased risk of malformations [30–32], a Norwegian group longitudinally assessed the development of children whose mothers had PCOS and used MT during pregnancy [33]. When compared to controls, children born from obese mothers using MT had increased head circumference, whereas the offspring of normal-weight mothers who also used MT during pregnancy had reduced length and weight. In a subsequent study, the same group showed that these children displayed slightly higher mean BMI, higher body fat percentage, and higher prevalence of obesity when compared to the placebo group at 5–10 years of follow-up [34]. Similar results were found in another study with a follow-up of 9 years that demonstrated greater weight and BMI in the children of mothers with GDM who used MT during pregnancy when comparing to children whose mothers used insulin [35]. However, the clinical relevance of these findings is questionable, since both groups had weight and BMI in the normal range.

Discussion

We have reviewed the current evidence regarding several fertility-related scenarios where MT is used in women with PCOS. For ovulation induction, the current literature suggests that MT alone may be useful; however, the greatest benefit occurs when MT is associated with other ovulation-inducing medications, mainly CC and gonadotropins. Although LSC can improve symptoms and metabolic parameters in women with PCOS, it is still unclear whether LSC can add benefit to MT in ovulation induction and pregnancy, since the current level of evidence is low. It is noteworthy that the level of evidence is low for most of the recommendations, thus future studies may change the indications for the use of MT in these scenarios.

The limited data regarding the use of MT before or during ovarian stimulation for IVF/ICSI in women with PCOS points towards a decrease in the risk of OHSS and no benefit for LBR. Since most studies utilized long GnRH-agonist protocol, there is even less evidence for the use of MT with short GnRH-antagonist protocol, which is the current standard protocol for women with PCOS due to the lower risk of OHSS when compared to GnRH-agonist protocols [36–38]. Moreover, the lack of benefit regarding LBR could be explained by the fact that ovarian hyperstimulation using high doses of gonadotropins in women with PCOS is so effective that the beneficial effects of MT would not be noticed. There was also a large variation of the dose regimen for MT, which ranged from 1000 to 2000 mg per day. This confounding factor may explain the variability of the RCTs results and should be addressed in future trials, since individualization of MT dose based on the patient’s phenotypes and metabolic profiles may provide better outcomes. Unfortunately, we found no new published or ongoing RCT evaluating the use of MT during IVF/ICSI research on clinicaltrials.gov. Thus, it may take a long time before we have strong evidence for or against MT in this setting.

The conflicting evidences about the effects of MT use during pregnancy, specifically regarding the incidence of pre-eclampsia, GDM, and gestational hypertension, suggest that the pathophysiology of IR in women with PCOS may differ from women with diabetes mellitus. Conversely, there is low-quality evidence that MT may exert a beneficial impact in the obstetric outcomes, which could be partly explained by its effects in the endometrium. Insulin resistance can negatively affect endometrial function by decreasing important factors to embryo-endometrium interaction such as glycodelin and type 1 insulin-like growth factor-binding protein [22, 23]. Changes in endometrial vascularization and transmembrane glucose transporters activity are also mechanisms that explain the IR-induced endometrial dysfunction [24]. These changes may be associated with the increased risk of obstetric and fetal complications found in pregnant women with PCOS. Metformin has been shown to improve endometrial function, increasing uterine vascularization and the production of important paracrine factors for the maintenance of pregnancy [39–41]. However, larger RCTs with longer follow-up period and controlled for confounder factors are needed to generate better quality evidence. An updated MA is currently being conducted and might clarify the real impact of MT use during pregnancy on GDM [42].

Concerning the risks of MT use during pregnancy, no evidence of teratogenicity has been found in the literature; however, there is limited evidence that MT can affect anthropometrics parameters of the offspring, with unknown long-term consequences. Therefore, the risk–benefit of continuing MT during pregnancy should be carefully assessed and individualized.

Table 1 summarizes the main findings of this review.

Table 1.

Summary of findings

Metformin is superior to placebo for OI during timed intercouse in women with PCOS

Metformin plus CC is superior to CC alone for OI during timed intercouse in women with PCOS

Metformin plus gonadotropins is superior to gonadotropins alone for OI during timed intercouse in women with CC-refractory PCOS

For OI in women with PCOS, metformin should be started up to 12 weeks before ovulation induction and continued until ovulation or pregnancy, at doses ranging from 850 to 2000 mg/day

Metformin use during ovarian stimulation for IVF/ICSI in women with PCOS decreases the risk of OHSS when GnRH-agonist protocols are used

During ovarian hyperstimulation in women with PCOS, metformin should be started up to 16 weeks before ovarian stimulation and continued at least until the day of the trigger, at doses ranging from 1000 mg/day to 2550 mg/day

It is unclear if metformin use during pregnancy decreases the risk of early pregnancy loss in women with PCOS

Metformin use in women with PCOS is associated with mild AE in up to 40% of cases. Prolonged-release presentations and progressive scaling up of the dose can decrease the incidence of AE

Metformin use during pregnancy at regular doses does not increase the risk of malformations

AE = adverse effects, CC clomiphene citrate, hCG humana chorionic gonadotropin, IVF/ICSI in vitro fertilization/intracytoplasmic sperm injection, OHSS ovarian hyperstimulation syndrome, OI ovulation induction, PCOS = polycystic ovary syndrome.

Conclusion

For women with PCOS, metformin is a good adjunctive medication for ovulation induction or hyperstimulation during IVF/ICSI, increasing success rates when combined with other medications. The adverse effects are mostly mild, and there is no risk of teratogenicity, but the risk of long-term complications for the offspring is not yet defined. In addition, the impact of MT use during pregnancy on the obstetric outcomes is unclear. The great heterogeneity of the studies limits the extrapolation of the findings, thus, more studies with larger populations, and evaluating specific phenotypes and dose regimens are needed to determine which group of women with PCOS benefit the most from the medication.

Author contribution

AGLN and FTLN conceived and implemented the study. All authors read and approved the final manuscript.

Declarations

Ethics approval

This study was exempted from the institutional review board approval, as it was a review.

Conflict of interest

The authors declare no competing interests.