Effects of acarbose on polycystic ovary syndrome: a meta-analysis

Whether or not acarbose benefits patients with polycystic ovary syndrome (PCOS) remains controversial. Therefore, we performed a meta-analysis to synthesize the literature regarding the therapeutic effects of acarbose on PCOS.

A comprehensive literature search was performed using terms, such as polycystic ovary syndrome, ovary polycystic disease, PCOS, hyperandrogenemia, acarbose, alpha-glucosidase inhibitors, and randomized controlled trials (RCTs), in the following bibliographic databases: Medline; Embase; and Cochrane Controlled Trials Register. The identified reference lists were checked manually.

6 RCTs met the inclusion criteria. Based on the meta-analysis of 3 studies, acarbose was superior to placebo or no treatment in reducing serum levels of testosterone (Std MD = - 3.38, 95% CI:-5.97~-0.78, P = 0.01) and acarbose caused a significantly higher incidence of side effects, such as abdominal distention and diarrhea (OR = 23.78, 95% CI: 5.67~99.75, P < 0.0001). The changes in Ferriman-Gallwey score or body mass index (BMI) were not significant. Based on the meta-analysis of 2 studies, acarbose were superior to placebo or no treatment in reducing triglycerides (TG; WMD = -18.18, 95% CI:-36.30~-0.06, P = 0.05) and very low-density lipoprotein (VLDL) cholesterol (WMD = - 6.49, 95% CI:-9.14~-3.84, P < 0.00001), and increasing high-density lipoprotein (HDL) cholesterol (WMD = 5.14, 95% CI:1.73~8.55, P = 0.003). There were no significant differences between acarbose and metformin with respect to improvements in ovulation rate, menstrual patterns, or changes in serum levels of testosterone, adverse events, or BMI. Heterogeneities were detected during the meta-analysis.

This is the first meta-analysis to evaluate the role that acarbose plays in the treatment of PCOS. The currently available data showed that acarbose can reduce testosterone, TG, and VLDL, and increase HDL. Acarbose caused a significantly higher incidence of gastrointestinal disturbance. Given the small RCTs and poor quality of RCTs included, these results are not conclusive. A large-scale, randomized controlled study is needed to ascertain this uncertainty.