Objectives: Elevated circulating androgens are risk factors for several chronic, metabolic and reproductive disorders. Metformin is an insulin-sensitizing agent that may lower androgen levels. To evaluate the effects of metformin on endogenous androgens and SHBG levels in women, we conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) comparing metformin with placebo or no treatment.
Data source: We used OVID to search MEDLINE, EMBASE and CENTRAL until March 2007.
Review methods: Two reviewers independently extracted data on methodological quality, participants, interventions and outcomes of interest. Our a priori primary outcome was post-treatment measurements. In a secondary analysis, we evaluated the difference between the pre- and post-treatment levels. We computed the weighted mean difference (WMD) as a measure of effect for each outcome using the DerSimonian-Laird random effects method. We used the I2 statistic to assess heterogeneity and explored its causes in subgroup analyses of features related to participants' characteristics and study design. Based on a regression model, we conducted sensitivity analyses by investigating the use of placebo as a predictor of effect size.
Results: Twenty RCTs fulfilled the inclusion criteria. Pooled WMDs in post-treatment levels between the metformin and control group were -0.31 nmol/l (95% CI -0.65 to 0.03) for total testosterone (TT), 0.10 pmol/l (95% CI -0.89 to 1.10) for free testosterone (FT), 0.14 micromol/l (95% CI -0.34 to 0.62) for dehydroepiandrosteronesulfate (DHEAS), -0.60 nmol/l (95% CI -1.67 to 0.46) for androstenedione (AND) and 5.88 nmol/l (95% CI 2.01-9.75) for SHBG. Pooled WMDs of the pre- to post-treatment differences (i.e. with adjustment for baseline hormone levels) were -0.38 (95% CI -0.51 to -0.25) for TT, -2.71 (95% CI -10.35 to 4.93) for FT, -0.50 (95% CI -0.83 to -0.16) for DHEAS, -1.39 (95% CI -2.30 to -0.49) for AND and 6.63 (95% CI 2.32-10.94) for SHBG. In subgroup analyses, features related to the administered treatment (i.e. metformin as a single agent or as part of combined regimens) partly explained the heterogeneity. Sensitivity analyses of studies using placebo showed similar results to those not using placebo.
Conclusions: Our systematic review and meta-analysis provides evidence of metformin-induced changes in circulating androgens and SHBG levels in women but the quality of evidence is not high. However, there are no data from RCTs regarding these effects in postmenopausal women or healthy premenopausal women. High-quality RCTs are required to evaluate whether metformin has effects on surrogate markers and patient-important outcomes in these patient groups.