Use of metformin to treat pregnant women with polycystic ovary syndrome (PregMet2): a randomised, double-blind, placebo-controlled trial

Background Women with polycystic ovary syndrome (PCOS) have an increased risk of pregnancy complications. Epi-analysis of two previous randomised controlled trials that compared metformin with placebo during pregnancy in women with PCOS showed a significant reduction in late miscarriages and preterm births in the metformin group. The aim of this third randomised trial (PregMet2) was to test the hypothesis that metformin prevents late miscarriage and preterm birth in women with PCOS. Methods PregMet2 was a randomised, placebo-controlled, double-blind, multicentre trial done at 14 hospitals in Norway, Sweden, and Iceland. Singleton pregnant women with PCOS aged 18–45 years were eligible for inclusion. After receiving information about the study at their first antenatal visit or from the internet, women signed up individually to participate in the study. Participants were randomly assigned (1:1) to receive metformin or placebo by computer-generated random numbers. Randomisation was in blocks of ten for each country and centre; the first block had a random size between one and ten to assure masking. Participants were assigned to receive oral metformin 500 mg twice daily or placebo during the first week of treatment, which increased to 1000 mg twice daily or placebo from week 2 until delivery. Placebo tablets and metformin tablets were identical and participants and study personnel were masked to treatment allocation. The primary outcome was the composite incidence of late miscarriage (between week 13 and week 22 and 6 days) and preterm birth (between week 23 and week 36 and 6 days), analysed in the intention-to-treat population. Secondary endpoints included the incidence of gestational diabetes, preeclampsia, pregnancy-induced hypertension, and admission of the neonate to the neonatal intensive care unit. We also did a post-hoc individual participant data analysis of pregnancy outcomes, pooling data from the two previous trials with the present study. The study was registered with ClinicalTrials.gov, number NCT01587378, and EudraCT, number 2011-002203-15. Findings The study took place between Oct 19, 2012, and Sept 1, 2017. We randomly assigned 487 women to metformin (n=244) or placebo (n=243). In the intention-to-treat analysis, our composite primary outcome of late miscarriage and preterm birth occurred in 12 (5%) of 238 women in the metformin group and 23 (10%) of 240 women in the placebo group (odds ratio [OR] 0·50, 95% CI 0·22–1·08; p=0·08). We found no significant differences for our secondary endpoints, including incidence of gestational diabetes (60 [25%] of 238 women in the metformin group vs 57 [24%] of 240 women in the placebo group; OR 1·09, 95% CI 0·69–1·66; p=0·75). We noted no substantial between-group differences in serious adverse events in either mothers or offspring, and no serious adverse events were considered drug-related by principal investigators. In the post-hoc pooled analysis of individual participant data from the present trial and two previous trials, 18 (5%) of 397 women had late miscarriage or preterm delivery in the metformin group compared with 40 (10%) of 399 women in the placebo group (OR 0·43, 95% CI 0·23–0·79; p=0·004). Interpretation In pregnant women with PCOS, metformin treatment from the late first trimester until delivery might reduce the risk of late miscarriage and preterm birth, but does not prevent gestational diabetes. Funding Research Council of Norway, Novo Nordisk Foundation, St Olav's University Hospital, and Norwegian University of Science and Technology.