Intended for healthcare professionals

Clinical Review State of the Art Review

Novel therapies for diabetes mellitus in pregnancy

BMJ 2018; 362 doi: (Published 16 July 2018) Cite this as: BMJ 2018;362:k2034
  1. Maisa N Feghali, assistant professor1,
  2. Christina M Scifres, associate professor2
  1. 1Department of Obstetrics, Gynecology and Reproductive Sciences, Magee-Womens Research Institute, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA
  2. 2Department of Obstetrics and Gynecology, University of Oklahoma College of Medicine, Oklahoma City, OK, USA
  1. Correspondence to: M Feghali maisafeghali{at}


Diabetes is a common complication of pregnancy, and the prevalence of all types of the disease is increasing worldwide. Diabetes in pregnancy is associated with short term and long term adverse effects for mother and child. The goal of treatment of diabetes in pregnancy is to minimize maternal and fetal adverse events related to hyperglycemia. Treatment options vary by type of diabetes, from a focus on lifestyle modifications in gestational diabetes to continuous glucose monitoring and insulin pumps in pregestational diabetes. Nevertheless, given the commonality of hyperglycemia, considerable overlap exists in the treatment of different types of diabetes in pregnancy. Also, despite ongoing research on treatment of diabetes in pregnancy for decades, changes in the characteristics of the patient population have highlighted the limited effectiveness of different therapies. Specifically, despite the co-occurrence of obesity and diabetes, treatment recommendations including glycemic targets are not altered in such cases and a single optimal treatment strategy for each type of diabetes in pregnancy does not seem to exist. Rather, the approach to treating pregnant women with diabetes likely needs to be individualized to maximize the short term and long term health of mother and child. This article will review recent clinical studies to summarize established treatment strategies and introduce novel therapies for diabetes in pregnancy.


Diabetes affects 6-9% of pregnancies with approximately 99% of women having gestational diabetes, 0.5% having type 2 diabetes, and 0.3% having type 1 diabetes.1234 In particular, the prevalence of gestational diabetes varies by population and diagnostic criteria.5 The prevalence of both gestational and pregestational diabetes is increasing, likely owing to the increasing obesity rates.2367 All types of diabetes are associated with increased risk for hypertensive disorders of pregnancy, excessive fetal growth, fetal demise, macrosomia, and neonatal morbidity, along with long term risks for obesity and diabetes in the offspring.89 Unique to pregestational diabetes is the risk for early pregnancy loss and congenital anomalies, which is directly linked to periconception glycemic control.

The goal of diabetes treatment in pregnancy is to minimize maternal and fetal adverse events related to hyperglycemia. Physiologic changes of pregnancy include progressive increases in insulin resistance, weight gain, and changes in body composition, and each of these changes may affect the pharmacologic properties of diabetes treatment (fig 1). Treatment options vary by type of diabetes, but given the considerable overlap we have organized our review by treatment, summarizing established treatment strategies and highlighting novel therapies for diabetes in pregnancy (table 1). We will focus on recent clinical studies that have evaluated various lifestyle and therapeutic options for women with diabetes in pregnancy.

Fig 1
Fig 1

Potential treatments for diabetes in pregnancy and their proposed mechanism of action. CPAP=continuous positive airway pressure

Table 1

Summary of treatments for diabetes in pregnancy

View this table:

Sources and selection criteria

We obtained the references for this review from various sources including PubMed,, and the Cochrane Database of Systematic Reviews (1990 to April 2018) by using the following terms: “diabetes in pregnancy”, “gestational diabetes”, “diet and diabetes in pregnancy”, “lifestyle modifications and diabetes in pregnancy”, “exercise and diabetes in pregnancy”, “obesity and diabetes in pregnancy”, “glyburide in pregnancy”, “metformin and pregnancy”, and “insulin and pregnancy”. We prioritized articles on the basis of study method (randomized controlled trials (RCTs) and meta-analyses over longitudinal observational studies, and cohort studies) and their date of publication. We included only full text, English language, peer reviewed publications. We used clinical guidelines from the American Congress of Obstetricians and Gynecologists (ACOG), the Royal College of Obstetricians and Gynaecologists (RCOG), the Royal Australian and New Zealand College of Obstetricians and Gynaecologists (RANZCOG), the International Association of the Diabetes and Pregnancy Study Groups (IADPSG), the National Institute for Health and Care Excellence (NICE), the Endocrine Society, and the American Diabetes Association (ADA). We prioritized the results of guidelines in this review for quality according to the method outlined by the United States Preventive Services Task Force.

Preconception care

In women with pregestational diabetes, preconception care includes optimizing glycemic control before conception with a hemoglobin A1c goal of 6.5% (48 mmol/mol) to minimize the risk of congenital anomalies (fig 2).1310 Preconception counseling visits can also include assessment for comorbidities and a review of drugs to discuss potential teratogenic effects. Preconception counseling may also be useful in women at high risk for gestational diabetes, including those who are overweight or obese, have a history of previous gestational diabetes, have a strong family history of diabetes, or have impaired glucose tolerance (fig 3). One challenge is that many pregnancies are unplanned, and novel interventions to encourage family planning efforts in women with diabetes are urgently needed.

Fig 2
Fig 2

Suggested treatment algorithm for pregestational diabetes in pregnancy. CGM=continuous glucose monitoring; CSII=continuous subcutaneous insulin infusion

Fig 3
Fig 3

Suggested treatment algorithm for gestational diabetes in pregnancy. GDM=gestational diabetes mellitus

Lifestyle modifications

Lifestyle modifications including diet, exercise, and weight management are first line treatment in women with gestational diabetes and important adjuncts to drug therapy in pregestational diabetes (fig 3). Weight gain recommendations are based on the Institute of Medicine’s guidelines for weight gain during pregnancy, which account for maternal body mass index but are unchanged by a diagnosis of diabetes in pregnancy.11 Exercise may improve glucose tolerance during pregnancy in women with gestational diabetes, and the ACOG and ADA have endorsed exercise as a helpful adjunct therapy in the treatment of diabetes.134

Between a third and a half of women with gestational diabetes are able to achieve glycemic control with diet alone,12 but the optimal diet in pregnant women with diabetes is debated. The ACOG and the Endocrine Society support a low carbohydrate diet, whereas the ADA and the Fifth International Workshop on Gestational Diabetes have withdrawn specific recommendations on diet or macronutrients for women with diabetes because of the absence of adequate RCTs.341314 A diet low in simple carbohydrates (33-40% of calories) is thought to limit postprandial glucose excursions and the associated risk of excessive fetal growth.15 However, this approach necessitates an increase in dietary fat,15 which may promote insulin resistance in humans and has resulted in adiposity, hepatic steatosis, and metabolic syndrome in offspring in animal models.16 Maternal triglyceride and free fatty acid concentrations are strong predictors of excess fetal fat accretion, so a low carbohydrate, higher fat diet may have unintended consequences on fetal health.171819

Recent RCTs in women with gestational diabetes suggest that a diet higher in complex carbohydrate and fiber, low in simple sugar (low glycemic index), and lower in saturated fat may be effective in blunting postprandial hyperglycemia, preventing worsened maternal insulin resistance and excess fetal growth.2021 In a recent crossover pilot study, 16 pregnant women with gestational diabetes were randomized to a higher complex carbohydrate (60%), lower fat (25%) diet or a conventional low carbohydrate (40% of calories), higher fat (45% of calories) diet at the time of diagnosis.22 The higher carbohydrate diet resulted in a slightly higher (by 6%; P=0.02), but well below current glycemic targets, area under the curve for 24 hour glucose. However, in a separate pilot study of 12 women with gestational diabetes who followed the complex carbohydrate and fiber diet for seven weeks, postprandial free fatty acids were 20% lower (P=0.06) and measures of maternal insulin resistance and infant adiposity were improved, although these differences were not sugnificant.23 These findings suggest a possible benefit of eating more complex carbohydrates and less fat. However, more definitive studies are needed to better define the optimal diet for pregnant women with diabetes, especially in obese women and in women with pregestational diabetes.

Glucose monitoring

Blood glucose monitoring is a cornerstone of diabetes management in pregnancy. Fasting and postprandial blood glucose monitoring is recommended in all type of diabetes, and preprandial testing is also recommended for pregnant women with pre-existing diabetes who are using insulin pumps or basal-bolus therapy.13410 Although HbA1c assessment is less burdensome than frequent daily testing, it is considered a secondary measure of glycemic control in pregnancy because A1c concentrations fall during normal pregnancy owing to red blood cell turnover and HbA1c does not reflect variability in glucose concentration.1310

Continuous glucose monitoring (CGM) technology was assessed in the CONCEPTT study, which randomized 325 pregnant women with type 1 diabetes to real time CGM or capillary glucose monitoring.24 Real time CGM users spent more time in target (68% v 61%; P=0.003) and had less hyperglycemia (27% v 32%; P=0.028) between randomization and 34 weeks’ gestation.24 CGM use was also associated with an approximately 50% reduction in large for gestational age (LGA) births (odds ratio 0.50, 95% confidence interval 0.28 to 0.90), admission to a neonatal intensive care unit (NICU) (0.48, 0.26 to 0.86), and neonatal hypoglycemia (0.45, 0.22 to 0.89).24

In addition, an RCT that assigned 340 women with gestational diabetes to either self monitored glucose or self monitored glucose plus intermittent CGM use found that CGM use was associated with lower mean birth weight (3138 (SD 484) v 3345 (508) g; P<0.001), less macrosomia (4.1% v 10.8%; P=0.03), less pre-eclampsia (3.4% v 10.1%; P=0.02), and fewer primary cesarean deliveries (34.7% v 46.6%; P=0.03).25 By comparison, a recent multicenter RCT found no significant difference in the risk of macrosomia between CGM and self monitored glucose in 300 pregnant women with gestational diabetes and pregestational diabetes.26

More recent trials published in the past two years have focused on devices that use flash glucose monitoring (FGM) technology, which can be worn for a period of time; users can obtain glucose measurements instantly by scanning the glucose sensor with the reader, producing real time data. Studies on the efficacy of FGM are limited to non-pregnant adults. Two RCTs, one including 328 adults with type 1 diabetes and another in 224 adults with type 2 diabetes, compared FGM with self monitoring of blood glucose for six months. Both found that FGM was associated with reduced rates of hypoglycemia (in type 1 diabetes, time in hypoglycemia reduced by 0.14 hours on average per day (P<0.001); in type 2 diabetes, by 0.70 hours on average per day P<0.001) and high patient satisfaction based on few adverse events attributed to the device.2728 However, in a pilot RCT in 40 adults with type 1 diabetes and a history of severe hypoglycemia or impaired awareness of glycemia, real time CGM use was associated with less time spent below 59.4 mg/dL (3.3 mmol/L) compared with FGM (2.4% v 6.8%; P=0.006).29

Studies on FGM use in pregnancy are limited to a single case report and a published abstract that noted good agreement between FGM and capillary glucose monitoring in 74 pregnant women (24 with type 1 diabetes, 11 with type 2 diabetes, and 39 with gestational diabetes).30 Further studies are needed to assess the utility of FGM in pregnancy, but this technology has appeal given the frequency of blood glucose monitoring in pregnancy.

Optimal glycemic targets

Fetal adiposity is strongly associated with elevated maternal glucose, and approximately 15-20% of pregnant women with diabetes whose glucose targets are within current clinical targets still deliver macrosomic infants who are at increased risk for long term metabolic dysfunction.313233

The Hyperglycemia and Pregnancy Outcomes (HAPO) study better defined the relation between maternal glycemia and fetal overgrowth, but current pregnancy glycemic targets have not been rigorously defined or tested in RCTs. In an observational study in pregnant women without diabetes, mean fasting (70.9 (SD 7.8) mg/dL (3.9 (0.4) mmol/L)) and one hour postprandial (108.9 (12.9) mg/dL (6.1 (0.7) mmol/L)) blood sugars as assessed by continuous glucose monitoring (CGM) were significantly lower than the current fasting (<95 mg/dL (5.3 mmol/L)) and one hour postprandial (<140 mg/dL (7.8 mmol/L)) glucose targets in pregnancies with diabetes, suggesting that current glycemic targets may be too high.34 In particular, obese women with gestational diabetes are at higher risk for adverse pregnancy outcomes than their normal weight counterparts,35 and obese women also have higher mean fasting and postprandial blood sugars despite higher doses and more frequent use of drugs.12 In a retrospective cohort of 1344 women with gestational diabetes, obese women who had fasting blood glucose above 88.7 mg/dL (4.9 mmol/L) or one hour postprandial blood glucose above 123.8 mg/dL (6.9 mmol/L) were at the highest risk for macrosomia, suggesting that tighter glycemic control could help to optimize outcomes in this high risk population.12 Lower glycemic targets may require more intensive dietary modification and both earlier and more aggressive dosing of drugs. This is the basis for ongoing studies such as the GDM-MOMS study, which is randomizing overweight and obese women with gestational diabetes to either standard glycemic targets (fasting <95 mg/dL (5.3 mmol/L), one hour postprandial <140 mg/dL (7.8 mmol/L)) or more intensive glycemic control (fasting <90 mg/dL (5 mmol/L), one hour postprandial <120 mg/dL (6.9 mmol/L)) and is expected to report in early 2009 (NCT02530866).


Insulin is the preferred treatment for pregestational diabetes in pregnancy, and several professional societies endorse it as a first line therapy in gestational diabetes.34 The type of insulin, timing of administration, and frequency are based on individual glycemic patterns. In women with pregestational diabetes, insulin is traditionally administered through multiple daily injections (MDI) combining short acting insulin for mealtime excursion and long acting insulin for basal glycemic control. The safety and efficacy of long acting analogs such as glargine and detemir have been demonstrated in pregnancy.36373839 Although commonly used, most insulin analogs have not been specifically tested in women with gestational diabetes, and few studies provide data specific to this population. A comprehensive discussion of different insulin formulations is beyond the scope of this review, and we will limit the discussion to an overview of established formulations followed by a focus on novel therapies. Data are lacking on newer insulin analogs, with no published reports on the use of Toujeo, Lispro-U200, or Basalgar in pregnancy at the time of writing and a single report of two pregnancies treated with degludec from the first trimester of pregnancy.40 No malformations were seen in the infants, but both needed admission to NICU due to hypoglycemia.40 Given the lack of data, use of newer insulin analogs in pregnancy should be further explored.

Recent technological advances have resulted in a programmable pump device that can manipulate the timing, quantity, and type of insulin through a continuous subcutaneous insulin infusion (CSII). These devices can be programmed to provide varying basal and bolus concentrations of insulin at differing times during the day, without abrupt changes and additional injections. Outside of pregnancy, the benefits of CSII compared with MDI include lower hemoglobin A1c levels, reduced variation and blood glucose concentrations, fewer hypoglycemic episodes, reduction in total daily insulin dose, and more flexibility of lifestyle.41 However, the value of CSII in improving pregnancy outcomes remains unclear.424344

A retrospective study assessed two comparable groups of women with type 1 diabetes, of whom 100 were using CSII and 44 were using MDI. Metabolic control determined by HbA1c measurements every trimester improved during pregnancy in both groups, but control was achieved earlier (second trimester compared with end of pregnancy) among participants using CSII.45 At parturition, patients using CSII had lower HbA1c levels (6.2% (SD 0.7%) v 6.5 (0.8%) (44 v 48 mmol/mol); P=0.02) and needed less insulin (P<0.01). However, maternal and neonatal outcomes did not differ.44 Similar findings of improved HbA1c levels with CSII versus MDI were reported in a case-control study in 99 pregnant women with diabetes, again with no differences in maternal or neonatal outcomes.46 A recent systematic review of six RCTs that compared the use of multiple daily insulin with CSII in 213 pregnant women found no significant differences between groups, including the frequency of maternal (odds ratio 1.35, 95% confidence interval 0.60 to 3.03) and neonatal hypoglycemia (1.31, 0.59 to 2.94) and the rates of preterm birth (1.29, 0.45 to 3.71), cesarean delivery (1.39, 0.76 to 2.55), stillbirth (2.5, 0.53 to 11.77), and LGA infants (1.04, 0.36 to 3.01).43

Subsequent studies have focused on the combination of CSII with CGM. The CONCEPTT study, described earlier, stratified women by insulin delivery (CSII or MDI) and found that CGM reduced hyperglycemia, limited glycemic variability, and improved HbA1c levels for CSII and MDI users.24 Also, sensor augmented insulin pumps have been developed to integrate real time CGM and insulin pump technologies. A proof of principle crossover trial was performed in 16 pregnant women with type 1 diabetes.47 Women were randomly assigned for a four week period to using a CSII pump and a continuous glucose sensor with or without tablet computer software that automatically adjusted the pump basal insulin rate at night. After four weeks, they switched to the other treatment group.47 Closed loop therapy resulted in a higher percentage of time in the designated euglycemic range (63-140 mg/dL (3.5-7.8 mmol/L)) than control therapy (74.7% v 59.5%; 95% confidence interval for the difference 6.1 to 24.2; P=0.002), and the overnight mean glucose concentration was lower with closed loop therapy than control therapy (119 v 133 mg/dL (6.6 v 7.4 mmol/L); P=0.009).47 After the initial eight week trial, 14 women continued to use the closed loop system for approximately 15 additional weeks, including time in labor and delivery. In these women, glucose concentrations were in the target range 68.7% of the time with a mean concentration of 126 mg/dL (7.0 mmol/L).47 However, no clear benefit of closed loop therapy on adverse pregnancy outcomes was seen, as 13 of the 16 newborns in the study cohort had a birth weight greater than the 90th centile.47

Two recent systematic reviews found no evidence to recommend one type of insulin or insulin regimen over another or continuous glucose monitoring over intermittent monitoring.4849 Although the findings from CONCEPTT are encouraging, and despite the advances in CSII and CGM, rates of adverse pregnancy outcomes remain high in women with type 1 diabetes. Further studies are needed to explore combining CSII and CGM with other treatment modalities to further reduce maternal and fetal risks.


Metformin is a biguanide that decreases hepatic glucose output, decreases intestinal glucose absorption, and increases peripheral glucose uptake in muscles and adipocyte cells.50 Metformin is a hydrophilic compound with low molecular weight and low binding capacity to plasma proteins, and trans-placental passage occurs in a dose dependent manner that is most likely carrier dependent.51 Observational studies have shown that concentrations of metformin in cord blood range from 50% to 100% of maternal concentrations, and fetal concentrations may even be higher than maternal concentrations.525354 Metformin does not seem to affect human placental glucose uptake or transport.55

Metformin to treat diabetes in pregnancy

Table 2 provides a summary of studies that have compared metformin with either insulin or glyburide. The MiG study remains the largest RCT comparing metformin plus insulin if needed with insulin alone, and this study has played an important role in recent recommendations that metformin can be considered as an acceptable alternative to insulin for treatment of gestational diabetes.56 We will discuss the data that led to these recommendations and the ongoing concerns about use of metformin in pregnancy.

Table 2

Summary of studies of metformin in pregnancy

View this table:

In the MiG trial, glycemic control was similar between the two groups, but 46% of women in the metformin group needed supplemental insulin.56 Rates of neonatal composite morbidity were similar between groups, but metformin treatment was associated with fewer instances of severe neonatal hypoglycemia.56 In contrast, metformin was associated with more preterm birth (12.1% v 7.6%; P=0.04).56 It was also associated with less weight gain between study enrollment and 36-37 weeks’ gestation (0.4 (SD 2.9) kg in the metformin group versus 2.0 (3.3) kg in the insulin group; P<0.001).56 Importantly, metformin had good patient acceptability, with 76.6% of women suggesting that they would choose metformin in a subsequent pregnancy compared with 27.2% of women assigned to insulin who would choose that.56

Several systematic reviews and meta-analyses have compared outcomes between women treated with metformin and either insulin or glyburide. These meta-analyses have reached disparate conclusions, likely owing to their varied approaches and whether unpublished studies or those also enrolling women with type 2 diabetes were included. The systematic review by Balsells et al found that metformin was associated with less maternal weight gain, improved postprandial glycemic control, and fewer cases of severe neonatal hypoglycemia but more preterm birth.57 In contrast, a recent network meta-analysis that included nine published and unpublished trials found that the risk of most outcomes, including LGA (relative risk 0.80, 95% confidence interval 0.64 to 0.99), macrosomia (0.75, 0.57 to 0.98), admission to NICU (0.74, 0.57 to 0.97), neonatal hypoglycemia (0.68, 0.50 to 0.92), and pre-eclampsia (0.56, 0.37 to 0.85), was lower in women randomized to metformin compared with insulin.58 There authors did not find a difference in preterm delivery (relative risk 1.37, 0.62 to 3.01) and concluded that metformin had the highest probability of being the most effective treatment when compared with insulin or glyburide.58 Another meta-analysis that included 16 studies of women with gestational diabetes or type 2 diabetes found no differences in the rates of preterm delivery (relative risk 1.18, 0.67 to 2.07).59 Although the most common focus in gestational diabetes is the risk of fetal overgrowth, the risk of increased preterm birth with metformin has not been resolved, and preterm birth is also associated with long term health consequences.6465 The high rate of failure with metformin has raised questions about appropriate dosing of metformin during pregnancy, and the increased renal clearance of metformin may necessitate dosing modifications.5366

Metformin use is common in gestational diabetes, but limited data exist regarding its use to treat type 2 diabetes in pregnancy. The study by Ainuddin and colleagues that used metformin to treat type 2 diabetes found that a significant proportion of women needed insulin, but metformin use was associated with less maternal insulin use and lower weight gain.61 Metformin was associated with fewer hypertensive disorders of pregnancy but more small for gestational age infants.61 An additional study by Ibrahim and colleagues of 90 women with gestational diabetes or type 2 diabetes found that the addition of metformin to insulin reduced the risk of neonatal hypoglycemia.60 Several ongoing RCTs comparing metformin with placebo in women with either type 2 diabetes or early onset gestational diabetes receiving insulin (Medical Optimization of Management of Type 2 Diabetes Complicating Pregnancy (MOMPOD), NCT02932475; and Metformin in Women With Type 2 Diabetes in Pregnancy Trial (MiTy), NCT01353391) will provide useful information on the effect of metformin as adjuvant therapy in this population, particularly with regard to the possible reduction in pre-eclampsia rates.67

Metformin use in women without gestational diabetes

Several studies in obese women without diabetes have explored the use of metformin to decrease the risk of fetal overgrowth. The EmPOWaR study by Chiswick and colleagues randomized 449 obese pregnant women with normal glucose tolerance to either metformin or placebo and found no significant difference in birthweight centile between groups.62 Metformin was also evaluated in a study of 400 obese women (body mass index >35) without diabetes who were randomized to metformin or placebo. Metformin did not affect birthweight Z scores, but it was associated with less maternal weight gain and a lower prevalence of pre-eclampsia.63 Metformin lowers soluble fms-like tyrosine kinase 1 and soluble endoglin secretion from primary human tissues. It may also reduce endothelial dysfunction, increase vasodilation, and induce angiogenesis, all of which show biologic plausibility that metformin could have the potential to prevent or treat pre-eclamspia.62

Potential long term effect of metformin on offspring

In addition to its glucose lowering properties, metformin inhibits proliferation of cancer cells by suppressing the production of mitochondrial dependent metabolic intermediates needed for cell growth, and it also causes down-modulation of cell proliferation related proteins through activation of AMPK and a decrease in mTOR.6869 These actions of metformin have raised interest in the potential short and long term effects of metformin on fetal and childhood development. In a follow-up of the MiG study, children aged 2 years whose mothers were treated with metformin had significantly higher mid-upper arm circumferences (17.2 (SD 1.5) v 16.7 (1.5) cm; P=0.002) and higher subscapular (6.3 (1.9) v 6.0 (1.7) mm; P=0.02) and biceps skinfolds (6.0 (1.9) v 5.6 (1.7) mm; P=0.04) compared with infants whose mothers were treated with insulin alone.70 However, no differences were seen in total fat mass, percentage body fat, or waist circumference. The authors speculated that metformin resulted in a more favorable fat redistribution with more peripheral fat and less visceral fat, but others have pointed out that only a subset of children underwent DEXA testing and there were no differences in waist circumference between groups.7071 Longer term follow-up data are pending for these children.

Another follow-up of 211 children from the MiG study found no significant differences in neurodevelopmental and psychomotor outcomes between those who were or were not exposed to metformin.72 Follow-up of a small Finnish RCT comparing metformin and insulin found that children exposed to metformin in pregnancy were significantly heavier (10.47 (SD 1.49) v 9.85 (1.26) kg; P=0.038) at 12 months and taller and heavier (12.05 (1.87) v 11.32 (1.45) kg; P=0.04) at 18 months despite having similar birth weights.73 No differences existed in motor, social, or linguistic development at 18 weeks between the two groups, although the study was small with only 93 children.73 Importantly, recently published follow-up data from 4 year old offspring of women with polycystic ovarian syndrome who were randomized to metformin or placebo showed that antenatal exposure to metformin resulted in higher weight and body mass index Z scores in the offspring, and there were more overweight and obese children in the group exposed to metformin.74


Sulfonylureas are insulin secretagogs that bind to the sulfonylurea receptor 1 (SUR1) of ATP sensitive K+ channels of pancreatic β cells, which causes channel closure and results in increased insulin secretion.75 Glyburide is hepatically metabolized and effluxed from the fetal to the maternal compartment against a concentration gradient by placental breast cancer resistance protein.76

Glyburide to treat diabetes in pregnancy

Glyburide is predominantly used to treat women with gestational diabetes. The largest trial evaluating use of glyburide randomized 404 women with gestational diabetes to insulin or glyburide and found similar glycemic control and neonatal outcomes for the two treatments (table 3).79 One additional benefit was that only 4% of the women who were treated with glyburide needed supplemental insulin. Although recent studies have shown that concentrations of glyburide in cord blood are 50-70% of maternal concentrations,7687 at the time of the above study glyburide transplacental transfer was thought to be minimal.7988 Likely as a result of these findings, glyburide treatment in women with gestational diabetes dramatically increased from 7% to 65% over the time period from 2000 to 2011.89

Table 3

Summary of studies of glyburide in pregnancy

View this table:

The increased prevalence of glyburide use has prompted multiple additional observational and randomized clinical trials comparing outcomes in women treated with glyburide with those treated with insulin. In one “natural experiment” that arose in a US healthcare system, the authors compared outcomes in 236 women treated with glyburide with those in 268 historical controls treated with insulin.77 More women in the glyburide group achieved treatment goals, and no differences were seen in birth weight or macrosomia.77 Glyburide was associated with higher rates of pre-eclampsia and need for phototherapy in infants, but fewer infants from the glyburide group were admitted to NICU.77 In contrast, a retrospective cohort study comparing outcomes in 4982 women treated with glyburide and 4191 women treated with insulin found that glyburide treatment was associated with increased risk of NICU admission, respiratory distress, hypoglycemia, birth injury, and LGA birth weight.78 This study was limited by the lack of information on glycemic control or maternal body mass index. However, these data raised important questions about whether the differences in outcomes between women treated with glyburide and insulin were related to unmeasured confounding or whether glyburide use outside of a small, tightly regulated clinical trial was associated with adverse perinatal outcomes due to either suboptimal glycemic control or direct effects of the drug.

Several recent meta-analyses have also looked at outcomes in women treated with glyburide compared with insulin.80838485 The meta-analysis by Balsells et al found that women treated with glyburide had higher birthweight infants, more macrosomia, and more neonatal hypoglycemia than those treated with insulin.57 The average rate of treatment failure among women treated with glyburide was 6.4%. Both the observational studies and meta-analyses of RCTs raise questions about the appropriate use of glyburide. The largest trial to date suggests that as long as glycemic control is comparable, outcomes are also similar between groups.79 The other trials included in these meta-analyses have small sample sizes and differ in their approach to glyburide dosing. In addition, the observational studies showing increased risk have methodological concerns including an inability to account for the level of glycemic control. Beyond gestational diabetes, data on the use of glyburide to treat type 2 diabetes in pregnancy are scarce, but this is one potential novel aspect of glyburide use. In one small cohort study, outcomes were similar between women treated with oral hypoglycemic agents (primarily glyburide) and insulin, with less weight gain and similar glycemic control between those receiving oral agents and insulin.90

Controversies in glyburide dosing

Several factors have been associated with a higher rate of failure of glyburide treatment, including a fasting plasma glucose above 110 mg/dL (6.1 mmol/L) on the oral glucose tolerance test, older maternal age, multiparity, and diagnosis of gestational diabetes before 25 weeks, but data to ensure that providers can select the most appropriate candidates or dosing regimens for glyburide use are limited.919293 Efficacy of glyburide in pregnancy may also be related to changes in drug metabolism, and the dosing used in some studies may not be ideal. During pregnancy, glyburide concentrations increase within 30-60 minutes, peak in two to three hours, and return to baseline by eight hours.87 These data identify several considerations for glyburide dosing. Firstly, glyburide concentrations return to baseline within eight to 10 hours of ingestion, so a pre-breakfast dose may not provide an adequate insulin response for dinner.94 All of the glyburide studies used either once or twice daily dosing, but this could lead to inadequate glycemic control because increasing the morning dose to control post-dinner blood sugars may result in hypoglycemia earlier in the day. Secondly, glyburide should be administered 30-60 minutes before a meal to ensure that glyburide concentrations increase before the rise in blood sugar seen after a meal.94 In addition, plasma concentrations of glyburide are lower in pregnancy than in the non-pregnancy state.87 Such findings suggest a possible need to change the dosing and administration schedule for glyburide in pregnant women with diabetes to maximize the efficacy of treatment.

Potential long term effect of glyburide on offspring

One important concern regarding glyburide use is the lack of long term follow-up data on children who were exposed to glyburide in utero, and these concerns have become more pressing with the knowledge that glyburide crosses the placenta. Recent data suggest that glyburide may also increase placental GLUT1 expression, which could augment fetal glucose delivery.95 Whether glyburide use increases the risk of fetal overgrowth independently of glycemic control is unclear, but LGA birth weight is thought to be a risk factor for adverse childhood metabolic outcomes.9697 Hypothetical concerns also exist about long term β cell function in offspring exposed to glyburide in utero, highlighting the importance of long term studies of growth and metabolism in offspring.

Glyburide versus metformin

Several small studies have compared glyburide with metformin. The meta-analysis by Balsells et al compared glyburide with metformin,57 but it included only two studies.8081 It found that metformin was associated with less maternal weight gain, lower birth weight, less macrosomia, and fewer LGA births.57 Preterm birth was increased in women who were treated with metformin. In contrast, a recent systematic review compared outcomes between women randomized to either glyburide or metformin from several additional studies and found no differences in maternal or neonatal outcomes.86 The authors noted that most of the studies comparing glyburide with metformin were of moderate to low quality, and they also stressed that the benefits and potential harms of one oral antidiabetic drug therapy compared with another are unclear.

Finally, a recent RCT of 104 women took a novel approach and randomized patients to either glyburide or metformin and then added the other drug if control was inadequate.82 Treatment failure after first line treatment because of poor glycemic control or adverse effects was similar between glyburide and metformin (34% v 29%; P=0.6), and treatment success after second line therapy was higher in the metformin group than in the glyburide group (87% v 50%; P=0.03).82 More patients in the glyburide group ultimately needed insulin compared with those in the metformin group (17% v 4%; P=0.03). No differences were seen in maternal weight gain, birth weight, LGA births, or macrosomia.82 However, a trend was seen toward more hypoglycemia in the metformin group. The authors concluded that the combination of glyburide and metformin may allow for a higher efficacy rate with a significantly reduced need for insulin, and they also postulated that the use of metformin first may have potentiated the effect of glyburide when the latter was added.82 However, further studies are needed to assess whether these drugs can be safely and efficaciously substituted for insulin and would require the combination to be directly compared with insulin alone. Using two drugs that cross the placenta to avoid insulin raises concerns about the potential long term programming effects of fetal hyperinsulinemia (glyburide) and altered hepatic gluconeogenesis, insulin sensitivity, mitochondrial function, and cell cycle proliferation (metformin).98

Alternate oral agents

α glucosidase inhibitors inhibit the α glucosidase enzymes present on the brush border of the small intestine, which slows carbohydrate absorption and reduces postprandial glucose concentrations. Only acarbose has been studied in pregnancy. One small case series found that postprandial glucose values normalized and infants were healthy in six women treated with acarbose three times a day.99 Another small RCT conducted in Brazil compared insulin versus glyburide versus acarbose in 70 patients.100 No significant differences in LGA and cesarean delivery rates among the three groups were seen, but acarbose was associated with more gastrointestinal side effects.100

Several other agents including DPP4 inhibitors, GLP-1 receptor agonists, SGLT-2 inhibitors, and thiazolidinediones are used outside of pregnancy, but minimal data are available in pregnancy. DPP-4 inhibitors reduce DPP-4 mediated degradation of endogenous incretin hormones. This leads to enhanced insulin synthesis and secretion as well as suppressed glucagon secretion. GLP-1 receptor agonists activate the GLP-1 receptor and enhance insulin synthesis and secretion in a glucose dependent fashion. SGLT-2 inhibitors inhibit activity of glucose transporters in the proximal tubule, serving to increase renal glucose excretion. Thiazolidinediones activate the PPAR family of nuclear receptors and reduce insulin resistance. In a mouse model, exposure to rosiglitazone resulted in abnormal placental morphology and altered expression of proteins implicated in placental development, raising concerns about the therapeutic use of this class of drugs during pregnancy.101

Emerging treatments

More than 200 studies are listed on under the heading of diabetes treatment in pregnancy. The studies primarily focus on glyburide, metformin, and insulin treatment to improve pregnancy outcomes, mostly in women with gestational diabetes. Nutrient supplements such as myo-inositol are also being tested to decrease the risk of adverse pregnancy outcomes in women with gestational diabetes. Myo-inositol is hypothesized to act through complex pathways that ultimately shift glucose intracellularly and then into fatty acids synthesis.102103 Myo-inositol supplementation starting at 12-13 weeks of pregnancy decreased the likelihood of gestational diabetes in at risk groups such as obese women and those with polycystic ovarian syndrome.104105106 Trials in women with other risk factors for gestational diabetes showed mixed results, with a recent trial showing no effect of myo-inositol supplementation in pregnant women with a family history of diabetes.107 Furthermore, evidence in support of myo-inositol as a treatment for gestational diabetes is weak,108 but it is the subject of a current investigation (NCT02149992).

Outside of pregnancy, obstructive sleep apnea (OSA) has been linked to enhanced inflammatory and oxidative stress responses, endothelial damage, and metabolic derangements.109110 Given the overlap between these biologic pathways and adverse pregnancy outcomes, including gestational diabetes, recent studies have focused on evaluating the link between OSA and adverse pregnancy outcomes. Rates of OSA increase with advancing gestation and may be even higher in women with gestational diabetes.111112 OSA was independently associated with gestational diabetes (odds ratio 3.47, 1.95 to 6.19) after adjustment for age, body mass index, chronic hypertension, and gestational weight gain.111 The study also noted an increasing exposure-response relation between OSA and gestational diabetes.111 Also, in women with gestational diabetes, short sleep duration was associated with worsened glucose control.113 These findings are the basis for a pilot study assessing the effect of continuous positive airway pressure treatment on glycemic control in gestational diabetes (NCT02245659).

Lastly, ongoing studies are exploring the relation between diabetes, obesity, and the gut microbiome. Evidence that the gut microbiota might influence obesity began with the observation that sterile, germ-free mice have decreased capacity for energy utilization compared with colonized counterparts.114 Data suggest that patients with type 2 diabetes have a reduction in the Firmicutes phylum, and the ratio of Bacteroides to Firmicutes correlates with plasma glucose concentrations.115 Kuang et al did a metagenome-wide association study comparing fecal samples collected at 21-29 weeks from 43 women with gestational diabetes and 81 healthy controls, and they found changes in microbial composition that could be used to identify women at risk of gestational diabetes.116 Others found that the placental microbiota and microbiome differed between women with and without gestational diabetes.117 The stool microbiota of insulin resistant women with a history of gestational diabetes has been characterized postpartum, and women with former gestational diabetes had relatively higher abundance of the Prevotellaceae family as well as reduced abundance of Firmicutes, similar to the situation in type 2 diabetes.118 Another cross sectional study of the gut microbiome in women who had gestational diabetes and their offspring five years postpartum found no significant differences in gut microbiota between women who had gestational diabetes and controls, but some differences existed between offspring of women who had gestational diabetes and those who did not.119

A few RCTs have examined the effects of probiotics administered during pregnancy with the aim of improving insulin sensitivity and reducing risk of gestational diabetes. In one RCT, pregnant women were randomized to either Bifidobacterium lactis alone or B lactis plus Lactobacillus rhamnosus GG (LGG) probiotic 14 days before a scheduled cesarean delivery.120 Both treatments decreased the expression of toll-like receptor genes in the placenta and meconium of neonates compared with placebo.120 A Finnish study randomized 256 women during the first trimester to dietary counseling and either a daily LGG and B lactis probiotic or placebo. Dietary counseling and probiotics decreased the frequency of gestational diabetes from 36% to 13%, but no significant differences in birth weight were seen.121 Of note, maternal central adiposity at six months postpartum was lower in women treated with diet and probiotics.121 In another RCT, 175 obese pregnant women were assigned to receive either daily Lactobacillus salivarius probiotic or placebo from 24 to 28 weeks’ gestation, and the probiotic did not affect maternal glycemia or birth weight.122


Guidelines for the management of diabetes in pregnancy include those from NICE, ACOG, ADA, and RANZCOG (table 4).13410123 All recommend treatment of all forms of diabetes in pregnancy to decrease adverse maternal and fetal outcomes. All agree that management of pregnancy complicated by diabetes should include medical nutritional therapy, counseling, and increased frequency of glucose monitoring. Differences exist in specific glucose targets, including the goal for fasting glucose below 90 mg/dL (5.0 mmol/L) by RANZCOG and a range of HbA1c goals of 6-7% for women with pregestational diabetes. Also, guidelines for treatment of gestational diabetes differ, specifically regarding the use of oral agents. Both SMFM and NICE have endorsed the use of metformin as an alternative to insulin for treatment of gestational diabetes.10124 However, we would argue that the available data on oral agents are limited, especially regarding long term effects after in utero exposure. Further research is needed to fully support the use of oral agents as alternates to insulin in gestational diabetes. All of the guidelines recognize that recommendations for the management of diabetes in pregnancy are largely based on studies with small sample sizes and less than level A evidence.

Table 4

Summary of guidelines on diabetes on pregnancy

View this table:


Diabetes in pregnancy has significant short and long term implications for affected mothers and their offspring. Despite decades of research, complications rates are high, particularly in pregestational diabetes, highlighting that novel interventions are urgently needed. Future studies are needed to clarify whether diets higher in complex carbohydrates and lower in fat are more effective than lower carbohydrate diets. Data on continuous subcutaneous insulin infusion suggest that this technology may improve outcomes in women with type 1 diabetes, and future studies are needed to assess whether it can improve outcomes in women with type 2 diabetes and even difficult to control gestational diabetes. Although oral agents are not novel, robust debate continues about the use of oral agents in gestational diabetes. We support insulin as first line treatment for gestational diabetes and hope that ongoing studies will clarify a potential role for metformin in the treatment of gestational diabetes and as an adjuvant to insulin in women with type 2 diabetes. We would also suggest that the approach to clinical trials in gestational diabetes would benefit from more novel approaches. Studies on treatment of diabetes in pregnancy have followed a traditional “simple parallel arm” clinical trial design, but gestational diabetes is a disease with a heterogeneous physiology characterized by various levels of insulin resistance and insulin secretion defects. Therapeutic studies outside of pregnancy recognize the heterogeneity of specific diseases and the need for more tailored treatment regimens, and these approaches could help us to better understand predictors of treatment failure. Also, future studies should engage patients to ensure access to novel therapies and understand their preferences regarding various treatment strategies.

Research questions

  • What diet in pregnant women with diabetes optimizes glycemic control and other metabolic characteristics and ultimately results in the most favorable fetal growth profile?

  • Are there different glycemic targets for subgroups of women with diabetes in pregnancy (for example, pregestational versus gestational diabetes or obese versus non-obese)?

  • Are there alternative metabolic or nutritional targets in pregnant women with diabetes to minimize maternal and fetal adverse outcomes?

  • What are the long term effects of diabetes treatment options, specifically oral hypoglycemic agents, on the offspring of pregnant women with diabetes?


  • Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors

  • Contributors: MNF and CMS both did the literature search, wrote the draft article, and revised the manuscript. They are both guarantors.

  • Funding: MNF is supported by the National Institutes of Health through grant number KL2 TR001856. The funding source had no involvement in the preparation, analysis, and interpretation of the data or submission of this review.

  • Competing interests: We have read and understood BMJ policy on declaration of interests and declare the following interests: none.

  • Provenance and peer review: Commissioned; externally peer reviewed.

  • Patient involvement: No patients were asked for input in the creation of this article.


View Abstract