Type 2 diabetes (T2D), once a disease primarily diagnosed in middle-aged and elderly adults, is increasingly affecting people of younger age groups [1]. With more women of childbearing age impacted, there has also been a corresponding rise in T2D prevalence in pregnancy in recent decades observed both in Europe and North America [2], [3], [4], [5], [6]. It is estimated that 21.1 million (16.7 %) of all live births globally in 2021 had some form of hyperglycemia in pregnancy, of which, 10.6 % were complicated by diabetes detected prior to pregnancy and 9.1 % were due to diabetes (either type 1 or type 2) first detected in pregnancy [1].

Youth-onset T2D prevalence and incidence vary by ethnicity and socioeconomic factors. The highest incidence of T2D in youth are reported in indigenous populations such as Canadian First Nations, American Indian and Navajo nation, Australian Aboriginal and African American populations, whereas youths from non-Hispanic Caucasian populations in Europe and the US have the lowest incidence rates [1]. Prior studies have also reported that T2D in pregnancy disproportionately affects those of racial-ethnic minorities and neighborhoods of lower socioeconomic status [7], [8].

Mothers with T2D in pregnancy and their offspring are at high risk for complications. Compared to non-diabetic pregnancies, mothers with T2D in pregnancy are more likely to have miscarriage, hypertensive disorders of pregnancy, caesarean section (CS) and their babies are more likely to be born preterm, have neonatal hypoglycemia, respiratory distress syndrome, hyperbilirubinemia, and admission to the neonatal intensive care unit (NICU) *[9], [10]. Compared to those with type 1 diabetes (T1D), women with T2D have been found to have higher rates of perinatal mortality and their babies are more likely to be small for gestational age (SGA), these complications in part attributed to an increased presence of baseline comorbidities such as obesity, chronic hypertension, and metabolic syndrome *[9], [11]. In addition, a study reporting pregnancy outcomes in a cohort of women with youth-onset T2D highlighted that these women are at particularly high risk, as 65 % of the participants experienced pregnancy complications, including pregnancy loss in 25 %,preterm birth in 33 %, with high rates of large for gestational age (LGA) also observed (27 %) [12].

To improve perinatal outcomes in T2D pregnancies, interventions including frequent self-monitoring of blood glucose, medical dietary therapy, and antihyperglycemic medications to achieve maternal euglycemia are regularly recommended [13]. Glycemic control can be challenging to achieve in this population as women with T2D already have insulin resistance at baseline, which is further aggravated by the effects of placental hormones [14]. As a result, women often require multiple daily injections with large doses of insulin to achieve glycemic targets, leading to increased weight gain and risk of hypoglycemia. Metformin, being an oral antihyperglycemic agent often used as first-line therapy for T2D outside of pregnancy, is therefore a logical therapeutic adjunct to add on or continue in this population, as it may aid glycemic control by improving insulin sensitivity without contributing to weight gain or hypoglycemia [15]. However, metformin freely crosses the placenta and there are concerns of long-term adverse effects of prenatal exposure [16], [17]. As a result, there remains no consensus currently on whether metformin should routinely be continued in T2D pregnancy as adjunct therapy to insulin, or whether it may supersede insulin as first-line pharmacotherapy to be initiated in T2D pregnancies, with the American Diabetes Association (ADA) guidelines recommending against it and the National Institute for Health and Care Excellence (NICE) guidelines supporting its use [13], [18].

Metformin is a biguanide medication which has been used for diabetes treatment since the 1950s [15]. It acts through multiple mechanisms, including improving insulin sensitivity, reducing hepatic glucose production, and increasing glucose utilization peripherally [19]. Metformin also stimulates glucagon-like-peptide-1 (GLP-1) release, which promotes insulin secretion [19]. At the molecular level, it activates the AMP-activated protein kinase pathway, which is a crucial player in maintaining cellular energy balance [19]. Furthermore, increased secretion of growth differentiating factor 15 (GDF15) by metformin can suppress appetite and promote weight loss [20]. There is also data metformin inhibits the mammalian target of rapamycin (mTOR) pathway for cellular nutrient transport, which may influence fetal growth and nutrient bioavailability [21].

Earliest reports of metformin use in pregnancy date back to the 1970s in South Africa [22]. Human physiologic studies have shown that metformin freely crosses the placenta, with concentrations in umbilical cord plasma at time of delivery as high as that found in the maternal circulation [23]. Meta-analyses of metformin exposure in the first trimester have not shown increased congenital malformations [24], *[25]. However, metformin use during pregnancy was limited until the MiG trial, a large randomized trial which investigated the use of metformin during pregnancy in women with gestational diabetes, further reviewed in the next section.

The Metformin in Gestational Diabetes (MiG) trial, published in 2008, was the first large-scale randomized controlled trial (RCT) to evaluate the role of metformin in treating hyperglycemia in pregnancy [26]. In the MiG trial, 751 women diagnosed with GDM between 20 and 33 weeks of gestation were randomized to open-label treatment with metformin (to a maximum dose of 2500 mg and with supplemental insulin if required) or insulin only [26]. The primary composite outcome of neonatal hypoglycemia, respiratory distress, need for phototherapy, birth trauma, 5-minute Apgar score less than 7, or prematurity did not differ between the treatment groups, and no treatment-related serious adverse events were observed [26]. Secondary outcomes included less neonatal severe hypoglycemia (<1.6 mmol/L), less maternal gestational weight gain (GWG), improved maternal weight loss postpartum, and better treatment acceptability in the metformin group compared to insulin group [26]. Although 46.3 % of women treated with metformin needed insulin initiation to achieve glycemic targets, their insulin requirements were lower than those in the insulin alone group [26]. Higher rates of preterm delivery was found in the metformin group in this study [26], but not in subsequent meta-analyses of similar RCTs, as detailed below.

Since the MiG trial, a number of other RCTs have been published comparing metformin with insulin in GDM pregnancies. Their findings have been summarized in several meta-analyses on this topic [27], *[28]. For the mother, metformin compared to insulin monotherapy resulted in no difference in glycemic control, but lower GWG, less hypertensive disorders of pregnancy, and less CS [27], *[28]. For the offspring, metformin use was associated with lower birthweight, lower risk of LGA, macrosomia, NICU admissions and hypoglycemia [27], *[28]. No difference was found in risk of SGA or preterm delivery between metformin and insulin treatment in gestational diabetes [27], *[28]. Research to date thus establishes metformin as a safe and efficacious alternative to insulin for the treatment of hyperglycemia in pregnancy in the context of GDM, in the short term.

Aside from its use in GDM, metformin has also been studied for use in pregnancy in women with polycystic ovary syndrome (PCOS) and non-diabetic obesity. In PCOS, although several earlier observational studies showed benefits of metformin in preventing pregnancy loss and development of GDM [29], [30], available RCTs to date have failed to replicate these findings. In the PregMet study, 257 women with PCOS were randomized to receive metformin or placebo from first trimester through the end of pregnancy [31]. The composite primary endpoint of preeclampsia, GDM, and preterm delivery did not differ between the treatment groups, and there was no difference in fetal birthweight between the groups [31]. The investigators did note that women in the metformin group had less late miscarriage and preterm delivery compared to those in the placebo group, though this finding did not reach statistical significance [31]. The PregMet2 study, published in 2019, set out to test this hypothesis on a further 487 women with PCOS randomized to metformin versus placebo [32]. While the primary outcome of late miscarriage and preterm delivery did not differ significantly between the treatment groups, a post-hoc pooled analysis using data from this trial as well as the original PregMet study and a smaller pilot study showed metformin halved the risk of late miscarriage and preterm delivery compared to placebo [32]. Neither of the studies showed a difference in GDM diagnosis between the treatment groups [31], [32]. A recent meta-analysis on RCTs of metformin use in PCOS pregnancies confirmed that metformin treatment is associated with reduction in preterm delivery but not maternal GDM diagnosis, maternal preeclampsia, or fetal birthweight [33].

Metformin has also been studied for use in pregnancy to improve outcomes for women with non-diabetic obesity in several RCTs. The MOP study was a RCT that evaluated 450 women with body mass index (BMI) ≥ 35 who received either metformin or placebo, in addition to lifestyle counselling, between 12 and 18 weeks of gestation to the end of pregnancy [34]. No difference was found in maternal incidence of GDM or fetal birthweight z-score, but the metformin group had lower maternal gestational weight gain and a lower incidence of preeclampsia [34]. Another RCT, EMPOWaR, studied 449 women with BMI ≥ 30, who received metformin or placebo from randomization at 12–16 weeks gestation to end of pregnancy and found that metformin had no significant effect on fetal birthweight, maternal GDM diagnosis, or maternal gestational weight gain [35]. The GRoW trial examined the use of metformin compared to placebo in addition to dietary interventions in both groups, in 524 women with a BMI ≥ 25 [36]. There was less weekly maternal gestational weight gain and less risk of CS in the metformin group, but no difference was found in risk of maternal GDM diagnosis, maternal preeclampsia, or fetal macrosomia, LGA, SGA, or preterm delivery [36].

A summary of pregnancy outcomes following metformin use during pregnancy in women withs GDM, PCOS, and non-diabetic overweight/obesity, based on select systematic reviews and meta-analyses of RCTs in each population, is provided in Table 1.

Up until recently, there were few clinical trials that solely examined the role of metformin for treatment of T2D in pregnancy. Table 2 provides a summary of available RCTs evaluating metformin use in pregnancy for T2D.

Hickman et al. and Refuerzo et al. were two small RCTs conducted in the United States, involving 31 and 25 women respectively, that randomized participants to metformin (plus insulin treatment if needed to achieve glycemic targets) versus insulin monotherapy for treatment of T2D in pregnancy [37], [38]. The primary outcomes in both trials were maternal glycemic control parameters (mean fasting glucose in Hickman et al. and trimester-specific HbA1c in Refuerzo et al.) and no difference was found between the treatment groups [37], [38]. Hickman et al. found that metformin treatment reduced maternal insulin requirement by an average 67 units per day at the end of pregnancy compared to insulin monotherapy, and that women in the metformin group were more likely to prefer the same treatment in a future pregnancy [37]. Refuerzo et al. found that while women in both groups achieved glycemic targets by the third trimester, all women in the metformin group achieved hemoglobin A1c < 7 % by the second trimester compared to only 82 % in the insulin group [38]. No difference was found in other maternal or fetal secondary outcomes including birthweight, macrosomia, NICU admission, neonatal hypoglycemia, or gestational age at delivery, though the studies were not adequately powered to detect a difference in these outcomes [37], [38].

Ainuddin et al. conducted a larger RCT in Pakistan, where 250 pregnant women with T2D were randomized to either metformin or insulin treatment in pregnancy [39]. In this study, women receiving metformin treatment required less insulin, gained less weight, and were less likely to develop hypertensive disorders of pregnancy and preeclampsia than those treated with insulin [39]. Babies in the metformin group were more likely to be SGA, though no differences were detected in fetal birthweight or LGA between the groups [39]. Offspring of women who received metformin and later also required insulin were less likely to have CS, neonatal hypoglycemia, or NICU admission compared to metformin only or insulin only groups [39]. No difference was found in preterm delivery [39].

It is worth noting that all three of the above trials excluded women with T2D who were already on insulin prior to study enrollment and those with diabetes complications [37], [38], [39]. Refuerzo et al. further excluded mothers with T2D duration > 10 years, and Hickman et al. included women with early GDM diagnosed before 20 weeks gestation [37], [39]. Thus, the study participants were likely an overall healthier group than those included in the later MiTy and MOMPOD trials.

The Metformin in Women with Type 2 Diabetes in Pregnancy Trial (MiTy) was the first large, multicenter, randomized, double-masked, placebo-controlled trial, involving 502 women with T2D in pregnancy in Canada and Australia, that evaluated whether the addition of metformin versus placebo, both added to insulin, would change the risk of a composite of serious neonatal outcomes including pregnancy loss, preterm birth, birth injury, respiratory distress syndrome, neonatal hypoglycemia and NICU admission [40]. In this study, metformin-treated women achieved better glycemic control, required less insulin by 45 units per day, gained less weight, and had fewer CS. Their babies were born with lower mean birthweight, were less frequently extremely LGA, less were over 4 kg and had reduced adiposity measures compared to those treated with placebo [40]. Metformin-treated babies were also found to be more likely SGA [40].

The underlying mechanism for increased SGA in the metformin group was felt to be multifactorial. It is possibly a direct effect of metformin affecting the mTOR pathway, which can influence nutrient delivery across the placenta [21]. It may also be an indirect effect mediated via reduced maternal gestational weight gain, improved glycemic control in pregnancy, and reduced insulin requirements [41]. Compared to prior studies of metformin use in GDM, which did not show increased SGA associated with metformin use, women with T2D may have underlying vasculopathies that predispose them to placental insufficiency and fetal growth restriction. In a secondary analysis of MiTy, investigators identified predictors for SGA among MiTy participants [41]. These included the baseline presence of chronic hypertension or nephropathy, and the use of metformin throughout pregnancy [41]. Interestingly, SGA infants in the metformin group were delivered significantly later than SGA infants in the placebo group (37 vs. 35 weeks), and, among infants with birthweight below the fifth centile, SGA infants in the metformin group had fewer adverse neonatal composite outcomes compared to those in the placebo group (33 % vs. 80 %) [41]. This suggests that, despite infants in the metformin group experiencing a leftward shift in birthweight distribution, there was no increase in adverse clinical outcomes. Overall, the absolute risk of SGA in women with chronic hypertension and/or nephropathy who used metformin was high (25 %), thus it is reasonable to be cautious in the use of metformin in these populations.

Another large RCT studying metformin use in T2D pregnancies was the Medical Optimization of Management of Overt Type 2 Diabetes in Pregnancy (MOMPOD) study. This was a multicenter, randomized, double-masked, placebo-controlled trial conducted at 17 clinical sites in the US. Women were eligible to participate if they had a diagnosis of T2D prior to pregnancy or GDM diagnosed early in pregnancy (before 23 weeks gestation), had a viable singleton pregnancy confirmed by ultrasound, and required insulin for glycemic management. Participants were randomly assigned to metformin or placebo treatment, from enrollment (11 weeks to before 23 weeks gestation), in addition to continuing insulin, throughout pregnancy [42]. The primary outcome was also a composite, this time including fetal or neonatal death, neonatal hypoglycemia, umbilical artery pH < 7.05, shoulder dystocia, hyperbilirubinemia, preterm birth, LGA infant, SGA infant, and birthweight < 2500 g [42]. The trial was halted early (at 75 % target enrollment) due to futility, and a total of 794 individuals who took at least one dose of the study drug were included in the intention-to-treat analysis [42]. The authors found that the composite adverse neonatal outcome did not differ between the treatment groups, nor did other prespecified secondary outcomes such as maternal insulin dose, maternal hypoglycemia, maternal excessive weight gain, hypertensive disorders of pregnancy, and CS [42]. In terms of the fetal outcomes, metformin-treated babies were born lighter by an average of 155 g and were less likely to be LGA compared to those in the placebo group [42]. No increase in SGA was found [42].

Comparing the two studies, both MiTy and MOMPOD demonstrated no difference in their respective composite neonatal adverse outcomes when metformin was added to insulin treatment for women with T2D in pregnancy *[40], *[42]. Both studies found metformin treatment reduced fetal birthweight and reduced either extreme LGA (in the case of MiTy) or LGA rates (in the case of MOMPOD) *[40], *[42]. However, there were differences. SGA was increased in MiTy offspring of metformin-treated women but this was not seen in MOMPOD’s metformin-treated participants *[40], *[42]. Also, while MiTy found that metformin-treated babies had less neonatal adiposity, these parameters did not differ among treatment groups in MOMPOD *[40], *[42].

The two trials also differed in findings of several important maternal secondary outcomes. In MiTy, metformin treatment was noted to have maternal benefits including improved glycemic control, less insulin requirements, less weight gain, and fewer CS [40]. These findings were not replicated in MOMPOD, which found no difference in the above outcomes between the metformin and placebo-treated groups [42]. Both trials found there to be no difference in maternal hypertensive disorders of pregnancy between treatment groups *[40], *[42].

The divergent findings of the MiTy and MOMPOD studies may in part be explained by differences in the study population. While both studies included a large proportion of patients from ethnic minorities (70 % were non-European origin in MiTy and 85 % were non-White in MOMPOD), the specific compositions within the ethnic minorities varied *[40], *[42]. In MOMPOD the largest represented racial group was Black or African American being 29 %, and 52 % of all participants identified as Hispanic ethnicity [42]. In MiTy only 15 % were of African origin and 2 % were of Hispanic origin, while South and East Asians made up a larger (27 %) proportion of the study population [40]. The phenotype of T2D may differ in these study populations. Prior research has suggested that T2D in people of African origin is more a problem of insulin resistance, while the disease in people of East Asian origin is a problem with deficient insulin secretory function [43]. A higher baseline insulin resistance may explain in part why metformin failed to provide the same maternal benefits in the MOMPOD study such as reducing maternal insulin requirements and reduced weight gain compared to placebo.

An individual patient meta-analysis is planned and will combine the MiTy and MOMPOD trial data with the goal of gaining more power to answer these clinically important questions regarding the use of metformin in women with T2D.

It is worth noting that, in addition to MiTy and MOMPOD, there was another observational study published recently (n = 2255) by Yland et al. that used data from two US health care claims databases (one publicly and one commercially insured) to emulate a randomized trial [44]. In this study, women with pregestational T2D who had at least 1 prescription dispensing record of both metformin and insulin during a baseline period of 180 days prior to last menstrual period (LMP) who continued to be pregnant beyond the second trimester were included, and maternal and fetal outcomes among those who continued metformin during pregnancy were compared to those who discontinued it beyond first trimester [44]. The authors found that a composite pregnancy outcome of preterm birth, birth injury, respiratory distress, neonatal hypoglycemia, and NICU admission did not differ between those who continued versus those who discontinued metformin [44]. In the commercially insured cohort, metformin continuation was associated with increased SGA, but this finding was not present in the publicly insured Medicaid cohort [44]. It is interesting to note that racial/ethnic makeup of the publicly insured cohort mimicked that of MOMPOD [42], with 32 % of participants identified as Black, 30 % Hispanic, and only 3 % Asian [44]. Racial/ethnic information was not reported in the commercially insured group to see if participant characteristics were more similar to those of the MiTy study.

One of the ongoing reasons for provider reluctance to continue or add metformin for T2D in pregnancy is the uncertain impact of in utero fetal exposure on long-term offspring adiposity. Several follow-up studies of RCTs of metformin use in pregnancy in women with GDM or PCOS have reported higher BMI and increased adiposity in children with exposure to metformin in utero compared to those without exposure [45], [46], while other follow-up studies found no difference between the exposed and unexposed groups [47]. This concern was further highlighted in a 2019 meta-analysis of RCTs of metformin use for GDM by Tarry-Atkins et al., which reported that metformin-exposed neonates were heavier as infants and had a higher BMI by mid-childhood, compared to those treated with insulin [48]. At the time, only a few follow-up studies of RCTs were available, which limited the authors’ ability to draw definitive conclusions. Furthermore, as the authors chose to include only RCTs of metformin use for GDM, it excluded studies of metformin use for non-GDM indications such as T2D, PCOS, and non-diabetic obesity.

To further explore the risk of long-term metabolic harm in offspring exposed to metformin in-utero, we conducted an updated systematic review and meta-analysis in 2024 using a pragmatic approach by including children of women with GDM, T2D, PCOS and non-diabetic obesity [49]. In addition to RCTs and their follow-up studies, we also included cohort studies. A total of 18 studies reporting on 7975 children exposed to metformin in utero and > 1 million children without metformin exposure in utero were included [49]. Fourteen of the studies were follow-up studies of 9 RCTs (4 conducted in mothers with GDM, 1 in T2D, 1 in PCOS, and 3 in obesity), and four studies were cohort studies conducted using information from administrative databases [49]. The oldest offspring age at follow-up was 11 years [49].

When all maternal indications for metformin were pooled, we found that there was no difference in the continuous outcomes of BMI z-score and BMI as well as categorical outcomes of overweight and obesity in children exposed to metformin compared to their peers at oldest age reported (mean age of follow-up 4.4 years and longest followed cohort to 11 years) [49]. When studies were stratified by age of offspring at follow-up, while metformin-exposed offspring were heavier (by an estimated weighted mean difference of 216 g in weight or 0.19 higher in BMI z-score) in infancy (1–3 years), the difference dissipated by mid to late childhood (3–6 years and 6–11 years) [49]. This suggests that intrauterine metformin exposure had a time-limited effect on offspring adiposity that dissipated by mid-childhood [49]. This was true for both metformin-exposed offspring of mothers with diabetes as well as for any diagnosis [49]. While this could be a reassuring finding, early growth acceleration and catch-up growth in infancy can be harmful as they have been associated with long-term risks of obesity and metabolic disease in adulthood [50]. Also, the weighted mean age in children included was only 4.4 years for metformin-exposed and 4.5 years for controls, leaving the possibility that the children have not been followed long enough to detect a difference [49]. Longer term follow-up studies, particularly beyond puberty, are needed to definitively address this concern.

When stratified by maternal diagnosis there was a significant subgroup effect indicating the effect of prenatal metformin exposure on offspring adiposity differed by maternal diagnosis [49]. Metformin exposure was not associated with an increase in childhood adiposity in women with diabetes (includes studies of GDM and T2D) or non-diabetes obesity but it was for offspring of mothers with PCOS [49]. The weighted mean difference for BMI was 0.88 kg/m [2] for the PCOS subgroup, thus unlikely to be a clinically significant difference. 49 We note that, among the follow-up studies included in the above meta-analysis, only the MiTy-Kids study [51] followed offspring of mothers with T2D and adiposity data is only available until age 24 months. Thus, there continues to be a gap in data on longterm impacts of metformin exposure in offspring of T2D pregnancies. Data from the ongoing MiTy Tykes study, which is following growth parameters of children aged 5–11 years of age of consenting MiTy mothers, along with offspring follow-up studies of MOMPOD and other RCTs of metformin use in T2D pregnancies, will help to address this gap.

Besides long-term adiposity measures, another area of interest is to determine whether in utero metformin exposure has any consequences to offspring neurodevelopmental outcomes. These outcomes were explored in a recent systematic review and meta-analysis by Gordon et al [52]. A total of 5 RCT follow-up studies and 2 cohort studies were included that compared neurodevelopmental outcomes among offsprings with and without prenatal metformin exposure, and the longest duration of follow-up was to age 14 [52]. Meta-analyses showed that metformin exposure was not associated with poor neurodevelopmental outcomes up to age of 14, including no evidence of global neurodevelopmental delay in infancy (<2 years) and early childhood (age 3–5), nor any difference in motor or cognitive outcomes in childhood and early adolescence (age 2–5 and age 2–14, respectively) compared to unexposed children [52].