Pregnancy Outcome After First-Trimester Exposure to Metformin: a Meta-Analysis

Gilbert C, Valois M, Koren G. Fertil Steril. 2006;86:658-663.

Metformin is a common oral diabetes medication that is also used off-label as an infertility treatment secondary to polycystic ovary syndrome (PCOS). PCOS is associated with insulin resistance and is the most common cause of anovulatory infertility, affecting an estimated 5% to 7% of women of reproductive age. Because the teratogenic effects of metformin have not been established, women are routinely advised to discontinue metformin at the diagnosis of pregnancy. However, irregular ovulation is a clinical sign of PCOS, so delay in the diagnosis of pregnancy is not unusual and results in inadvertent, prolonged fetal exposure to metformin. This study sought to quantify the risk of spontaneous abortion, stillbirth, major malformations, minor anomalies, intrauterine growth retardation, and preterm labor among women with either type 2 diabetes or PCOS who were treated with metformin at least through the first trimester.

Only studies with a suitable control group were included in the meta-analysis. Review articles, letters to the editor, and animal studies were excluded. These conditions unexpectedly restricted the analysis to just 8 of 73 candidate studies, yielding a total of 172 treated patients and 235 control patients from 5 retrospective and 3 prospective controlled studies. Due to a lack of data, the only pregnancy outcome that could be analyzed was the risk of major malformations.

In the combined analysis, 1.7% (3/172) of treated patients and 7.2% (17/235) of control patients bore a child with a major malformation. The weighted odds ratio (OR) was 0.50 (95% CI, 0.15–1.60) with treatment. In the analysis of patients with diabetes, 0% (0/28) of treated patients and 7.6% (13/172) of control patients bore a child with a major malformation; the weighted OR was 0.85 (95% CI, 0.14–5.11) with treatment. In the analysis of patients with PCOS, 2.2% (3/139) of treated patients and 6.3% (4/63) of control patients bore a child with a major malformation; the weighted OR was 0.33 (95% CI, 0.07–1.56) with treatment.

A significant limitation of this meta-analysis not mentioned by the authors is that a single PCOS study in which the coauthors participated was weighted most heavily in the calculation of the OR. This study favored metformin treatment more strongly than any other study included in the meta-analysis. Unlike the other included studies, it has only been published as a meeting abstract. According to the complex formula used to compute the OR, it was given 44.44% of the statistical weight in the combined analysis and 77.42% of the statistical weight in the PCOS analysis.

The mean major-malformation rate in the control group was 7.2%, which is within the expected range for offspring of women with diabetes, and about 2 to 3 times higher than the background rate. The mean major-malformation rate in the metformin-treated patients was 1.7%, which is within the background rate. These data suggest that first-trimester metformin exposure does not increase the risk of major malformations, but further studies are needed to address the potential effects of metformin on spontaneous abortion, stillbirth, minor anomalies, intrauterine growth retardation, and/or preterm labor.