In our study, the prevalence of GDM among pregnant women in Qatar is 31.0%, which is similar to 31.6% of recent reported data from Qatar [1]. Women on metformin were found to be older and heavier, with a higher incidence of obesity at baseline, which could have contributed to their need for pharmacotherapy to achieve glycemic control.
Although the age of our cohort (31.9 ± 5.0 years) was similar to that previously studied in this population (32.5 ± 5.4 years) [10], the pre-pregnancy weight and BMI were lower compared to that cohort. The obesity rate in Bashir’s cohort was 59.8% compared to 30.6% in this study. The significant difference in this baseline characteristic between the cohorts could be attributed to the ethnic composition of the two samples. The proportion of Qatari women was higher in our cohort (59.3% vs. 41.8%), and the non-Qatari women hail from 51 countries.
In our cohort, the gestational age at delivery in the metformin-treatment group (37.8 weeks) was significantly lower than the diet-controlled group (38.6 weeks), which is similar to previously published data [10]. However, this cannot be explained by the increased induction of labor in the metformin group, as there was no difference in induction rates between the two groups. A plausible explanation for the lower gestational age at delivery in patients treated with metformin could be the increased incidence of polyhydramnios and LGA in this group.
Univariate regression analyses revealed an increase in the risk of C-section, polyhydramnios, and LGA in metformin-treated patients compared to the diet-controlled group. Previous studies have shown that age, pre-pregnancy weight, fasting glucose at diagnosis, and gestational weight gain were independent risk factors for various maternal and neonatal outcomes [26,27,28]. Therefore, we performed a multivariate analysis to minimize the effect of these confounding variables on our results. Our adjusted analysis found that patients treated with metformin were not at increased risk of adverse pregnancy outcomes.
In this study, the overall rate of C-Sect. (45.0%) among patients with GDM was similar to that (46.1%) reported by a retrospective cohort study in 2019 [10]. This study, also based in Qatar, compared GDM patients on diet and those on pharmacotherapy and showed that C-section was more likely in the treatment group (51.4 vs. 39.4%), and they had a higher rate of C-section (OR: 1.37, p <.001). Our C-section rates are also significantly higher (50.7 vs. 42.2%) in the metformin group compared to women treated with diet intervention.
In most studies, treatment with metformin was not associated with an increased incidence of cesarean section in women with GDM [29]. However, Ijäs et al. found that metformin-treated patients had a higher rate of C-Sect. (38% vs. 20%) than those treated with insulin (OR: 1.9, p =.047) [30]. The study suggested that the increased C-section rate was a result of prolonged labor or presumed fetal compromise. However, there was no difference in the incidence of emergency C-sections between our two study groups. Another cohort study of patients with GDM in Sweden from 2012–2016 found that metformin use alone was not associated with delivery by C-section compared to insulin use, but metformin and insulin use was associated with higher odds of C-section (OR: 1.65, 95% CI: 1.06–2.56), especially unplanned C-section in nulliparous women [31]. Women treated with both metformin and insulin were older, more often multiparous, and had higher BMI at baseline, which suggests that they were more likely to have worse hyperglycemia and more severe GDM [30]. These findings are replicated in the metformin group of our cohort and likely contributed to the increased incidence of C-sections compared to the diet control group. As these women have more risk factors for GDM, the need for the use of metformin suggests that the pregnancy is at higher risk and could be used as an early indicator that C-section may be required, allowing clinicians to plan for it.
The higher rate of overall C-sections in women on metformin could also be explained by the higher rates of LGA and polyhydramnios and higher mean weight percentile at birth in this group. The higher incidence of C-sections among patients on metformin cannot be attributed to the higher incidence of previous C-sections, as the rate of prior C-sections, although non-significant, is lower (65.4 vs. 72.4%) within this group. However, after adjusting for confounding variables, there was no significant increase in the risk of C-section between the two groups. The non-significant increase in risk after correction likely suggests that metformin is not the cause for the higher incidence of C-sections in women treated with it. It is recommended to consider an elective C-section in cases of suspected macrosomia with an estimated fetal weight exceeding 4,500 g in patients with GDM to avoid birth trauma-related injuries such as shoulder dystocia [32, 33]. However, no significant difference is seen in the rates of elective C-sections as the incidence of macrosomia is similar between the two groups in our study.
In addition, this study also looked at the frequency of assisted vaginal delivery between the two groups, and it was seen that it was significantly lower in the metformin-treated group. Perhaps this could be due to higher neonatal birth weight percentile and increased incidence of LGA in the metformin group. Thus, a C-section would have been the safer option to prevent shoulder dystocia and associated complications, such as brachial plexus injury in neonates born to women on metformin. As a result of more C-sections among patients treated with metformin, there were fewer assisted vaginal deliveries (6.2%) in this group. Conversely, there were more assisted vaginal deliveries (11.4%) in patients treated with diet control than those treated with metformin. Nevertheless, multivariate analysis showed no significant reduction in the risk of assisted vaginal delivery among patients on metformin.
Our study showed that there was no significant increase in the risk of preeclampsia among women treated with metformin compared to those on diet control. In contrast, some studies have reported that metformin use is associated with less gestational hypertension [34, 35] and preeclampsia [36] compared to insulin. The proposed mechanisms by which metformin prevents gestational hypertension and preeclampsia include moderating the production of antiangiogenic factors such as soluble vascular endothelial growth factor receptor-1 (sVEGFR-1) and soluble endoglin, mitigating endothelial dysfunction through its effect on the mitochondria and reducing the production of reactive oxidative species [16]. Metformin can also modify cellular homeostasis and energy disposition through rapamycin [16]. It is, however, recommended by the American Diabetes Association to avoid metformin in pregnancies complicated by hypertension, preeclampsia, or at risk for intrauterine growth restriction [37, 38]. As the effect of metformin on the mammalian target of the rapamycin (mTOR) signaling pathway could cause nutrient restriction to the fetus and placenta [39, 40], there is potential for acidosis or fetal growth restriction, especially in the setting of placental insufficiency [41]. Other proposed mechanisms include suppressing the mitochondrial respiratory chain, inhibiting folate-related metabolic pathways, and shifting cellular metabolism to a catabolic state [42, 43]. Metformin’s negative effect on fetal body composition and intrauterine growth has been studied preclinically in mice [44]. A retrospective study examining the use of oral antihyperglycemic drugs in 118 diabetic pregnancies between 1966 and 1984 concluded that treatment with metformin was associated with an increased prevalence of preeclampsia (32 vs. 10 vs. 7%, p <.001) compared to treatment with insulin or sulphonylurea [45]. However, clinical studies have mostly shown that metformin is either neutral or protective against the development of gestational hypertension and preeclampsia. On the contrary, LGA and neonatal birth weight percentile incidence is significantly higher in the metformin-treated group, which likely refutes the hypothesis of nutrient restriction and growth inhibition caused by metformin. Nonetheless, this “thrifty hypothesis”, which suggests that the in-utero environment created by metformin use predisposes to cardiometabolic disease onset in adulthood, needs to be further studied [46].
The increased incidence of polyhydramnios in women on treatment with metformin could be linked to more severe hyperglycemia as uncontrolled GDM is associated with polyhydramnios, and a possible explanation for its pathogenesis is osmotic diuresis because of fetal hyperglycemia leading to polyuria [49]. There is evidence demonstrating a strong association with high HbA1c through pregnancy in patients with polyhydramnios [47, 48]. Our results align with previous studies showing no association between metformin use and polyhydramnios risk in pregnancy [10, 49, 50]. However, it should be noted that the NICE guidelines recommend insulin use from the start, alone or in addition to metformin, in cases of fasting blood glucose between 108 and 125 mg/dL with concomitant polyhydramnios [4].
No difference was seen in the incidence of SGA between the two groups in this study. The increased risk of SGA has been extensively described in type 2 diabetes treated with metformin in pregnancy (MiTy trial) but not in GDM [51]. A large register-based cohort (CLUE) study in Finland found that, compared to insulin use, maternal exposure to metformin was associated with an increased risk of SGA [17]. It has been previously suggested that metformin may impact the risk of SGA through reduced maternal food intake, inhibition of folate-related pathways, and inhibition of the mammalian target of rapamycin (mTOR) [51]. In contrast, a prospective observational study of 60 women conducted in India found that the incidence of SGA, although non-significant, was lower in the metformin group compared to the insulin group [52]. However, this was not observed in our cohort, which agrees with findings from many previous studies [10, 21, 51,52,53,54,55,56]. Two systematic reviews and meta-analyses, including 12 and 19 studies, respectively, have reported no difference in the risk of SGA between the metformin and insulin-treated groups [57, 58]. In 2022, a follow-up to the CLUE study revealed that when compared with non-pharmacological treatment or drug naïve GDM patients, exposure to metformin was not associated with an increase in the risk of SGA in offspring of patients with GDM (OR 0.97, 95% CI 0.73 to 1.27) which aligns with our study findings [59].
LGA incidence was higher in the metformin group than in the diet control group, which could be explained by the severity of the disease. More severe maternal hyperglycemia implies that more glucose crosses the placenta, causing fetal hyperinsulinemia and ultimately large for gestational age (LGA) neonates. Combination treatment of metformin and insulin was associated with an increased risk of LGA in the Finnish register-based cohort [17]. However, only 10.9% of the treatment group in this study supplemented their metformin use with insulin. Therefore, it is less likely a consequence of the pharmacologic agents used and probably a consequence of their increased disease severity. Even so, there was no increase in the risk of LGA among the metformin-treated group after adjusting for age, pre-pregnancy weight, gestational weight gain, and fasting glucose.
Neonatal jaundice was also more frequent among the women on metformin, which replicates the findings of previous studies [10, 60]. However, the proportion of neonatal jaundice in the metformin group (1.9%) was far less in our cohort. It is suspected that its increased incidence is due to the relatively lower gestational age at birth, increased maternal hyperglycemia, and severity of GDM in the metformin group. Still, it should be noted that there was no increased risk of neonatal jaundice compared to the diet group when corrected for confounding factors.
The strengths of this study include the use of a single centre with consistent application of national guidelines for the screening and treatment of GDM. Screening women at their first antenatal care visit with FBG and HbA1c ensured that those with pre-gestational diabetes were excluded from the study. To our knowledge, this is the first study conducted in the high-risk MENA region, with a predetermined objective of comparing the impact of metformin use on maternal and neonatal outcomes to non-pharmacologic treatment or diet control alone. The previous study published in 2019 investigated outcomes between GDM patients on pharmacotherapy, including insulin alone, and those on diet [10]. On the other hand, this study excluded patients on insulin only and thus is better designed to answer the question of metformin’s short-term effect on maternal and neonatal outcomes in women with GDM.
Limitations
This study has several limitations. Its retrospective design limits causal interpretation and may introduce bias. Important confounding factors such as physical activity, smoking status, and adherence to treatment were not captured, though they may significantly influence maternal and neonatal outcomes. Additionally, data on glycemic control during pregnancy were not collected, which limits the ability to assess the full impact of treatment approaches.
We also did not have data on chronic hypertension and renal dysfunction, both of which could affect outcomes and should be considered in future research. Furthermore, we did not collect data on adherence to pharmacologic treatment, which may influence the effectiveness of therapy and subsequent outcomes. Finally, the metformin group may have included women with more severe gestational diabetes, introducing potential bias. Randomized controlled trials are needed to confirm and expand upon these findings.
Future direction
Considering the increasing prevalence of GDM in the Qatar population, metformin has become a popular choice for pharmacotherapy due to its oral route and convenience of use compared to insulin. However, there is a need to further study its impact, both short and long-term, on maternal and neonatal outcomes. There are a limited number of studies globally that compare neonatal and maternal outcomes in GDM between patients on metformin and drug-naïve patients. A prospective cohort study comparing neonatal and maternal outcomes between patients on metformin and diet control would be ideal to address the question of short-term safety of exposure to metformin in pregnancy. Examining outcomes between these two groups will not only validate the results of this study but also provide more accurate information on the adverse outcomes of metformin in pregnancy and the perinatal period. Studies on the long-term metabolic and neurodevelopmental outcomes in offspring exposed to metformin in utero are more limited, and therefore, more work is required to establish the long-term safety of metformin as an oral hypoglycemic agent in pregnancy.