These authors have contributed equally to this work
This article was submitted to Obstetric and Pediatric Pharmacology, a section of the journal Frontiers in Pharmacology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Polycystic ovary syndrome (PCOS), also known as Stein–Leventhal syndrome, was originally described by Stein and Leventhal in 1935. It is characterized by amenorrhea or occasional menometrorrhagia, hirsutism, infertility, and large, pale, polycystic ovaries with thickened capsules (
Nevertheless, some patients with PCOS show CC resistance (CCR), defined as failure to achieve ovulation after the dose of CC has been gradually increased to 150 or 250 mg/day—the final dosage differed among the studies—in at least three consecutive cycles (
Prominently, LOD is an invasive procedure and carries inherent risks. Conversely, drug safety is always the center of attention. Metformin has been widely used clinically for >60 years and there is sufficient evidence of its safety and tolerability in most populations (
In the present study, we conducted a systematic review and meta-analysis of published randomized controlled trials (RCTs) to evaluate the effectiveness and safety of Met, Met-CC, LOD, and LOD combined with CC (LOD-CC), as well as to compare Met/Met-CC (Met with or without CC) with LOD/LOD-CC (LOD with or without CC) in patients with CCR-PCOS who showed anovulation-related infertility. In so doing, we sought to establish which approach should be used as the primary treatment and to confirm which of the above three viewpoints is correct.
The present meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement protocol (
PubMed, Embase, and the Cochrane Central Register of RCTs were searched for articles published between 1 Jan 1966 and 31 Aug 2019. In accordance with the PICOS strategy, search terms were formulated as follows: 1) patient population: “polycystic ovarian syndrome”, “polycystic ovary syndrome”, “polycystic ovarian disease”, “PCOS”, and “PCOD”; 2) intervention: “metformin”, “dimethylguanylguanidine”, “dimethylbiguanidine,” “glucophage”, “dimethylbiguanide”, and “DMBG”; 3) control: “laparoscopic ovarian drilling”, “laparoscopic ovarian diathermy”, “LOD”, “laparoscopic ovarian electrocautery”, and “LOE”; 4) outcomes: “ovulation”, “pregnancy”, and “live birth”; 5) study design: “randomized controlled trial”, “random allocation”. Results were restricted to humans and no language restrictions were applied (
Two reviewers (ML Sun and L Zheng) independently screened the titles and abstracts to determine whether the articles were relevant to the meta-analysis based on the pre-defined inclusion criteria listed below. The full texts of potentially eligible studies were then reviewed before the final selection. Any disagreements were resolved in consultation with the principal investigator (XH Wang).
The inclusion criteria were as follows: 1) study population of patients with both CCR-PCOS and anovulation-related infertility; 2) intervention of LOD-controlled Met treatment despite continuing CC; 3) reporting of fertility outcomes (ovulation, pregnancy, and live-birth rates); 4) RCT study design. The exclusion criteria were as follows: 1) duplicates; 2) study design other than RCT (e.g., reviews, meta-analyses, case reports, guidelines, trial protocols); 3) absence of comparison between Met/Met-CC and LOD/LOD-CC; 4) absence of fertility outcomes.
Data were extracted from the included articles by two independent reviewers (M-LS and LZ) using standardized data extraction sheets. Any disagreements were resolved in consultation with the principal investigator (X-HW). If available, the following information was extracted from each article: study period, inclusion, and exclusion criteria, first author, year of publication, subjects’ country of residence, definition of CCR, number of participants in each group, clinical characteristics of the participants, treatment regimens, duration of treatment, follow-up period, and fertility outcomes (ovulation, pregnancy, and live-birth rates). Adverse events (AEs), including miscarriage, multiple pregnancy, ectopic pregnancy, ovarian hyperstimulation syndrome (OHSS), drug-related AEs, and intra- or post-operative complications, were identified according to the data originally documented in the articles.
Two researchers (M-LS and LZ) independently conducted quality assessment of all the included articles using the Cochrane risk-of-bias tool in the following domains: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias (
The overall meta-analysis was performed following the appropriate Cochrane Guidelines (
Heterogeneity among the included studies was analyzed using the
We conducted sensitivity analyses of the primary effectiveness outcome (live birth/ongoing pregnancy), to determine whether the conclusions were robust to arbitrary decisions made about eligibility, and analysis. Sensitivity analysis was performed using the “random-effects model” and “leave-one-out” methods.
If any heterogeneity occurred, the reasons for it were ascertained using a meta-regression analysis of the primary effectiveness outcome (live birth/ongoing pregnancy).
Subgroup analysis of the primary effectiveness outcome (live birth/ongoing pregnancy) was also performed to explain expected significant heterogeneity.
To reduce reporting bias, we were alert to data duplication and ensured that our search for eligible studies was comprehensive. Potential publication bias was examined using the funnel plot, Egger linear regression, and Begg rank correlation tests (
We generated a “Summary of findings” table using GRADEPro 3.6 software. This table shows the overall quality of the evidence for the main review outcomes according to the GRADE criteria. We justified judgements about evidence quality (high, moderate, low, or very low), and documented and incorporated these judgements into the reporting of results for each outcome.
The statistical significance level was set at
Overall, 71 studies were retrieved from the electronic databases. Twenty-three duplicates were removed, resulting in 48 unique titles. Following title and abstract review, we assessed eight full-text articles for eligibility. Five RCTs (
Preferred Reporting Items for Systematic Reviews and Meta-analyses flow chart depicting the process of paper selection and the number of papers in each phase. Notes: RCT, randomized controlled trial.
The characteristics of the included RCTs are summarized in
Characteristics of the included randomized controlled trials.
Study, authors (yr) | Study country | Treatment | Treatment regimens# | Duration of treatment& | The follow-up period | Frequency of intercourse (advice from researcher) | Time of pregnancy after treatment |
---|---|---|---|---|---|---|---|
Hashim HA. (2011) | Egypt | 1) Met-CC | 1) Met 500 mg tid for 6–8 weeks; then CC 100 mg for 5 days starting from day 3 of menstruation; increased by 50 mg for the next cycle. 2) LOD operation. 3) HCG was given when one follicle measuring at least 18 mm was found | 1) 6 cycles | 1) six cycles. 2) 6 months. 3) until the end of the pregnancy | Patients with ovulation to have intercourse naturally or 24–36 h after hCG injection | NR |
2) LOD | 2) one LOD | NR | |||||
|
Egypt | 1) Met | 1) Met 850 mg bid after dL. 2) LOD operation | 1) 6 cycles or 30 weeks | 1) six cycles or 30 weeks 2) 6 months. 3) until 13 weeks of gestation | To have Intercourse every other day when a mature follicle was seen on ultrasound | NR |
2) LOD | 2) one LOD | NR | |||||
|
Italy | 1) Met-CC | 1) Met 850 mg qd started from day 1 of a progesterone-induced withdrawal bleeding, and the dosage was increased after 1 week up to 850 mg bid. 2) LOD operation | 1) 6 cycles | 1) six cycles. 2) six cycles. (3) until the end of the pregnancy | To have Intercourse once per 3 days starting on day 9 after uterine bleedings | NR |
(2) LOD | 2) one LOD | NR | |||||
|
Italy | 1) Met-CC | 1) Met 850 mg bid after dL; placebo 1 tablet bid after LOD. 2) CC 150 mg qd for 5 days (the 3rd - 7th day of progesterone-induced uterine bleeding) | 1) 6 months | 1) 6 months. 2) 6 months. (3) until the end of the pregnancy | To have inter-course 4 times/2 days from the time of a follicle with Ø ≥18 mm | 7 (range 1–15) months$ |
2) LOD-CC | 2) one LOD | 6 (range 3–13) months$ | |||||
|
Italy | 1) dl-Met | Met 850 mg bid after dL; Multivitamins 1 tablet bid after LOD. | 1) 6 months 2) one LOD | 1) After the study, a 6-months extension of the follow-up period was done for each treatment group. 2) until the end of the pregnancy | To have inter-course/2 days for 4 times from the time of a follicle with Ø ≥18 mm | NR |
2) LOD-placebo | NR |
Study, authors (yr) | Groups | Definition of CCR (max dose of CC, mg/day) | Diagnosis of ovulation (progesterone level, ng/ml) | Full analysis set sample size (n) | Per protocol set sample size (n) | Clinical characteristics of participants | Fertility outcome (Rate, %) | |||
---|---|---|---|---|---|---|---|---|---|---|
Age of patients in each group (yrs) | BMI (kg/m2) | Ovulation | Pregnancy | Live-birth or ongoing pregnancy* | ||||||
Hashim HA, et al. (2011) | 1) Met-CC | 150 | 5 | 138 | 138 | 27.2 ± 2.5 | 26.2 ± 3.4 | 68.8 (95/138) | 64.5 (89/138) | 58.7 (81/138) |
2) LOD | 144 | 144 | 26.5 ± 2.3 | 26.1 ± 3.5 | 71.5 (103/144) | 66.0 (95/144) | 59.7 (86/144) | |||
|
1) Met | 150 | 5 | 55 | 55 | 23.6 ± 2.6 | 35.6 ± 4.4 | 58.2 (32/55) | 20.0@ (11/55) | 16.4* (9/55) |
2) LOD | 55 | 55 | 24.3 ± 4.5 | 36.1 ± 3.6 | 76.4 (42/55) | 38.2 (21/55) | 30.9* (17/55) | |||
|
1) Met-CC | 250 | 10 | 25 | 23 | 27.5 ± 4.8 | 30.2 ± 3.4 | 80 (20/25) | 56.0 (14/25) | 48.0 (12/25) |
2) LOD | 25 | 24 | 28.2 ± 4.3 | 29.8 ± 3.2 | 68 (17/25) | 60.0 (15/25) | 52.0 (13/25) | |||
|
1) Met-CC | 150 | 10 | 8 | 8 | 27.2 ± 2.2 | 28.4 ± 1.6 | 87.5 (7/8) | 75.0 (6/8) | 50.0 (4/8) |
2) LOD-CC | 20 | 20 | 25.4 ± 2.4 | 27.7 ± 1.8 | 70 (14/20) | 60.0 (12/20) | 35.0 (7/20) | |||
|
1) dl-Met | 150 | 10 | 60 | 54 | 26.8 ± 2.2 | 28.1 ± 1.7 | 70 (42/60) | 18.57 (39/60) | 53.3 (32/60) |
2) LOD-placebo | 60 | 55 | 27.5 ± 2.4 | 27.6 ± 1.6 | 58.3 (35/60) | 13.42 (31/60) | 33.3 (20/60) |
|
|
|
|
|
|
|
|
Hashim HA. (2011) | 1) Met-CC | 2.9 (4/138) | NR | 5.8 (8/138) | 0/138 | (1) 9.4% (13/138) of patients in Met-CC group suffered gastrointestinal side effects (mainly nausea and vomiting), but continued therapy. (2) LODs were completed successfully without intra- or post-operative complications | NR |
2) LOD | 0/144 | NR | 6.3 (9/144) | 0/144 | NR | ||
|
1) Met | NR | NR | 3.6@ (2/55) | NR/55 | (1) 18.2% (10/55) of patients in Met group reported self-limiting dizziness, nausea, and flatulence, but continued therapy. (2) LODs were completed successfully without surgical complications | NR |
2) LOD | NR | NR | 7.3@ (4/55) | NR/55 | NR | ||
|
1) Met-CC | 0/25 | NR | 8 (2/25) | NR/25 | (1) 17.4% (4/23) of patients in Met-CC group reported diarrhea, abdominal pain, hot flashes, and nausea, but continued therapy. (2) LOD was not performed in 1 obese woman. (3) 8% (2/25) cases: a minimal endometriosis and avascular pelvic adhesions | NR |
2) LOD | 0/25 | NR | 8 (2/25) | NR/25 | NR | ||
|
1) Met-CC | 0/8 | NR | 25 (2/8) | NR/8 | (1) Minor gastrointestinal symptoms, with no difference in AEs between 2 groups (12.5% vs 15.0%, |
NR |
2) LOD-CC | 0/20 | NR | 25 (5/20) | NR/20 | NR | ||
|
1) dl-Met | 0/60 | NR | 10 (6/60) | NR/60 | (1) 22.2% (12/54) patients in dL-Met group reported diarrhea, flatulence, and nausea; (2) 5.5% (3/55) in LOD-placebo group reported gastralgia and constipation. (3) No intra- or post-operative complications | 50.16€ for Met |
2) LOD-placebo | 0/60 | NR | 15 (9/60) | NR/60 | 1050 € for LOD |
Notes: Met, metformin; CC, clomiphene citrate; LOD, laparoscopic ovarian drilling/diathermy; dL, diagnostic laparoscopy; AEs, adverse events; SAE, serious adverse events. tid, thrice daily; bid, twice daily; qd, once daily. HCG, human chorionic gonadotropin injection. P, progesterone. yr, year. NR, not reported; OHSS, ovarian hyperstimulation syndrome. #, all treatments were suspended in patients who conceived. &, Those who conceived were observed until the end of the pregnancy (for up to further 9 months) to obtain live birth data for each treatment arm, except the study of Hamed HO, et al. (2010). *, ongoing pregnancy rate was calculated based on first trimester abortion. $, median time. @, first trimester miscarriage rate.
The included trials had a total of 590 subjects (from 28 to 282 in each study), of which 115 were treated using Met, 171 using Met-CC, 284 using LOD, and 20 using LOD-CC (
The results of the quality assessment are shown in
Risk of bias graph for reviewing authors’ judgements:
To compare Met/Met-CC with LOD/LOD-CC in terms of both effectiveness on fertility outcomes and safety, we pooled data from all subjects in the included trials using the fixed-effects model.
Effectiveness did not differ significantly between Met/Met-CC and LOD/LOD-CC in terms of live birth/ongoing pregnancy (RR = 1.02, 95% confidence interval [CI]: 0.87–1.21, z = 0.28;
Forest plot of effectiveness and safety outcomes in the Met/Met-CC vs LOD/LOD-CC comparison using the fixed-effect model:
Safety did not differ significantly between Met/Met-CC and LOD/LOD-CC in terms of miscarriage (RR = 0.79, 95% CI: 0.46–1.36, z = 0.86,
The
The
Pooling based on a random-effects model (M-H heterogeneity) resulted in similar live-birth/ongoing pregnancy rates (RR = 1.04, 95% CI: 0.76–1.42, z = 0.25,
We assessed the treatment methods, follow-up period after pregnancy, subjects’ country of residence, and samples using logistic meta-regression analysis. The results implied that different follow-up periods contributed to 15.43% of the heterogeneity (
The included studies were stratified by treatment methods. No significant difference occurred between Met-CC and LOD (RR = 0.97, 95% CI: 0.81–1.17,
The included studies were then divided into subgroups according to follow-up period. After excluding the study by Hamed et al., in which the pregnancy was only followed-up until 13 weeks’ gestation to obtain first trimester abortion and ongoing pregnancy data, no significant effect was found (RR = 1.09, 95% CI: 0.93–1.29,
Next, the participants were divided into subgroups based on the participants’ countries. No significant difference occurred between any of the treatment methods in either Egypt (RR = 0.91, 95% CI: 0.75–1.10,
The RCT published by Hashim HA (2011), which included the largest number of patients and cited by two reviews (
The funnel plot in
Funnel plot of
Egger linear regression and Begg rank correlation.
Fertility | Egger linear Regression | Begg rank correlation | ||||||
---|---|---|---|---|---|---|---|---|
Coef | 95% CI | t |
|
Kendall score (P-Q) | SD# |
|
|
|
Ovulation-induction rate | 1.322 | −4.373–7.017 | 0.74 | 0.514 | 6 | 4.08 | 1.22 | 0.221 |
Pregnancy rate | −6.109 | −21.33–9.111 | −1.28 | 0.291 | −6 | 4.08 | 1.22 | 0.221 |
Live-birth/ongoing pregnancy rate | −8.233 | −22.78–6.317 | −1.80 | 0.170 | −4 | 4.08 | 0.73 | 0.462 |
Miscarriage rate | 6.962 | −168.7–182.7 | 0.13 | 0.908 | 2 | 4.08 | 0.24 | 0.806 |
Note: #, SD, standard deviation of score; *,
Although drug-related AEs occurred at rates between 5.5% and 22.2%; the symptoms were self-limiting and did not affect treatment continuation. LOD was completed successfully without any intra- or post-operative complications in all but three patients: one with obesity who failed the laparoscopy and two who showed minimal endometriosis and avascular pelvic adhesions (
We found no evidence of a difference between Met/Met-CC and LOD, as well as between Met-CC and LOD-CC in terms of live birth/ongoing pregnancy, pregnancy, ovulation induction, and miscarriage rates. The “summary of findings” table showed that the quality of evidence was moderate in the Met-CC vs. LOD comparison, low for the pregnancy and ovulation induction rates of the Met vs. LOD comparison, and very low for the Met-CC vs. LOD-CC comparison, live birth/ongoing pregnancy, and miscarriage rates of the Met vs. LOD comparison (
In patients with anovulatory, infertility-related CCR-PCOS, our main analysis including all studies indicated no significant difference in effectiveness or safety between Met/Met-CC and LOD, as well as between Met-CC and LOD-CC. The robustness of this result was evaluated using sensitivity analysis. The quality of evidence was very low to moderate.
Moderate-quality evidence suggested that Met-CC and LOD should be recommended as parallel treatments for CCR-PCOS. The live-birth rates associated with Met-CC and LOD were 57.06% and 58.58%, respectively. The miscarriage rate was similar between the two (6.13% vs. 6.51%). The
Very-low quality evidence indicated that there was uncertainty about the effectiveness and safety of Met and LOD in terms of live birth/ongoing pregnancy and miscarriage rates. Similarly, the effectiveness and safety of Met-CC and LOD-CC were unclear in terms of live birth, pregnancy, ovulation induction, and miscarriage rates.
The results of Yu et al. (2017) showed no difference in live birth, pregnancy, ovulation induction, or miscarriage rates between Met/Met-CC and LOD (
Bordewijk et al. (2020) concluded that LOD/LOD-CC medical ovulation induction may lead to lower live birth rates in women with anovulatory CCR-PCOS than medical ovulation induction alone (
The strengths of the present systematic review included the extensive search strategy, as well as the meta-regression, subgroup, and sensitivity analyses. All included studies were RCTs with no publication bias, and the robustness of their results was confirmed by sensitivity analysis.
However, the present meta-analysis also had several limitations. For example, the number of included RCTs and participants was limited. In addition, the treatment methods were potential source of heterogeneity. The two RCTs comparing Met with LOD were highly heterogeneous, while the one comparing Met-CC with LOD-CC had a small sample size. No RCT compared Met with LOD-CC. Because the quality of evidence was low or very low, no conclusion could be drawn regarding the comparison of Met with LOD or of Met-CC with LOD-CC. Lastly, variation in the follow-up periods after pregnancy may have impacted fertility results. That said, the two groups in the trials were subject to the same experimental conditions in terms of follow-up time, which may have reduced heterogeneity.
Although moderate-quality evidence suggested that Met-CC and LOD should be recommended as parallel, second-line therapies for patients with CCR-PCOS, the present study indicates that Met-CC should be recommended as the optimum treatment for patients with CCR-PCOS because it is cheap, safe, and different from LOD, whose effect depends on operator proficiency; LOD intervention might be the first choice for patients with CCR-PCOS if they are willing to undergo diagnostic laparoscopy.
Very low-to low-quality evidence has suggested that there is little or no difference in effectiveness or safety between Met and LOD, as well as between Met-CC and Met-CC in women with CCR-PCOS, but our results showed that Met is a promising drug that could play a more important role in CCR-PCOS treatment because it is effective, safe, and cheap. Specifically, Met is an old drug with new applications, including PCOS treatment (
LOD reduces both testosterone and the luteinizing hormone/follicle-stimulating hormone ratio in women with PCOS, and it improves clinical outcome (
Additional large RCTs with adequate blinding are needed to more precisely estimate the difference between Met/Met-CC and LOD/LOD-CC. Such trials should comprehensively evaluate outcomes, including live birth, pregnancy, ovulation induction, AEs (multiple pregnancy, miscarriage, ectopic pregnancy, OHSS, drug-related AEs, and surgical complications), costs, patient satisfaction, long-term benefits (spontaneous resumption of ovulation and menstruation), as well as the potential risks of LOD (such as premature ovarian failure).
In conclusion, there is no evidence on the difference in the outcomes between the two interventions regarding ovulation, pregnancy, and live birth. As LOD is an invasive procedure and carries inherent risks, the use of metformin with or without clomiphene should be the second-line treatment for women with polycystic ovary syndrome (PCOS) who do not ovulate only by clomiphene citrate.
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/
All authors took part in the research design. M-LS and LZ performed the article screening and data extraction. M-LS wrote the first and successive versions of the manuscript. W-PB, H-YW and Q-KS contributed many constructive opinions and suggestions. All authors contributed to the interpretation of the results, intellectual content, critical revisions of the drafts of the paper, and approved the final version of the paper. As the principal investigators, X-HW, LZ, and G-LG had full access to all data in the study and had final responsibility for the decision to submit for publication.
This research was supported by the Open Research Funding of Central Laboratory, Beijing Shijitan Hospital Affiliated to Capital Medical University (Grant No. 2020-KF28). This work was also supported in part by the National Science and Technology Major Project for Significant New Drugs Creation (Grant No. 2017ZX09304026). The funders of the study had no role in data collection, data analysis, data interpretation, or writing of the article.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
We would like to thank Yong Wang for his contributions to formulating the search string of the systematic literature search.
The Supplementary Material for this article can be found online at: