Gestational diabetes mellitus (GDM) is carbohydrate intolerance with onset or initial recognition during pregnancy.[1] More than three decades since the original published description of GDM, no consensus exists regarding its implications or management. Paralleling the continuum of carbohydrate intolerance in the diabetes spectrum, researchers and practitioners demonstrate a continuum of response to GDM. At one end of the continuum, there exists an advocacy for aggressive glycemic control during pregnancy. Supporters of this approach note a successful reduction in macrosomia. At the other end of the continuum, there exists a nonaggressive (if not casual) approach to GDM; this position is bolstered by the lack of evidence-based research to substantiate the significant preventable morbidity from GDM. These divergent responses to GDM also represent extremes in the spectrum of resource utilization. This article critically evaluates the variety of treatment strategies utilized in the management of GDM.

A discussion of management strategies is worthwhile only if GDM is accepted as a disease entity with significant potential morbidity. This has been demonstrated most clearly in a cohort of women with diagnosed but untreated GDM.[2] The offspring of these women had a remarkably increased risk of macrosomia (44%) and birth trauma (25%). This report is representative of the GDM body of literature, demonstrating macrosomia as the primary morbidity associated with GDM. Secondary morbidities, which may be related to macrosomia, include fetal birth trauma, maternal birth trauma (including an increased risk for cesarean delivery) and neonatal hypoglycemia.[2,3] The continuum of carbohydrate intolerance is proportionally associated with a spectrum of severity for some of these morbidities. Among women with carbohydrate intolerance below the diagnostic threshold for GDM, the incidence of macrosomia (neonatal birth weight ≥4.5kg) and cesarean delivery is higher than for women without diabetes.[4] Furthermore, when stratified by 3-hour 75g oral glucose tolerance test result, a graded increase in the frequency of shoulder dystocia, neonatal birth weight ≥4kg and/or ≥90 percentile, and maternal hypertension is present among those without diabetes.[5]

Although GDM is the most significant risk factor for macrosomia, this condition is not unique to GDM. Indeed, the majority of neonates with macrosomia are born to women without diabetes. Attempts to limit macrosomia in the offspring of mothers with GDM will likely be hampered by the multi-factorial risks for macrosomia, especially maternal obesity (which is commonly present in women with GDM).

GDM also has multiple risk factors, including age over 25 years, family or personal history of diabetes, prior macrosomia, and prior fetal death. Women belonging to several ethnic groups have an increased risk of GDM (and type 2 diabetes), including those with Hispanic, African, Native American, South or East Asian, or Pacific Island ancestry. As an example of a disease supporting the ‘fetal origins’ hypothesis, susceptibility to GDM may be programmed in utero. Paralleling the inverse relationship of birth weight and later risk for type 2 diabetes, GDM has been demonstrated to be more common in women who had a birth weight of <3kg.[6] Thus, maternal obesity remains as the only remediable risk factor for GDM.

Progressive insulin resistance is a normal physiologic consequence of pregnancy, correlating with elevated levels of antagonistic hormones in the maternal circulation, including human placental lactogen, prolactin, estrogen, progesterone, and cortisol. Additionally, oxytocinase, histaminase, and alkaline phosphatase enzymatic activity are elevated during pregnancy, increasing the effective ‘insulinase’ activity. Furthermore, maternal adiposity, increased caloric intake, and decreased physical activity may contribute to the insulin resistance which develops during pregnancy. These resistance factors combine to increase the insulin requirement 3- to 4-fold during pregnancy. No obvious insulin receptor defect appears to account for either the insulin resistance of normal pregnancy or for GDM.

In pregnant women without diabetes, this insulin resistance is compensated for by an increased rate of insulin output, especially following meals.[7] Thus, despite progressive insulin resistance, the normal range (upper 95% CI) for capillary glucose in women without diabetes who consume an unrestricted diet remains ≤67 mg/dL fasting, ≤115 mg/dL 1 hour after meals, and ≤108 mg/dL 2 hours after meals.[8] Even for these women who do not have diabetes, the limited postprandial elevations in plasma glucose play an important role in fetal growth; postprandial plasma glucose levels correlate with fetal abdominal circumference.[8]

In pregnant women with GDM, the fasting insulin level is similar to or higher than that of those without diabetes, but not high enough to produce normoglycemia.[9,10] Additionally, the insulin response to a meal is delayed, with the peak plasma insulin level occurring approximately 90 minutes after the meal, rather than by 60 minutes postconsumption as in women without diabetes.[10] The relative ineffectiveness of insulin in GDM has also been demonstrated with the euglycemic insulin clamp (standardizing the insulin dose to body surface area, with the glucose infusion adjusted to maintain plasma glucose concentration at approximately 88 mg/dL). Both before the test and at steady state during the euglycemic insulin clamp test, women with GDM have a higher glucose production rate, lower glucose clearance rate, and a higher plasma free fatty acid concentration.[10] Placentas in pregnancies complicated by GDM have neither a change in facilitative glucose transporter (GLUT) 1 expression or glucose transport to account for the lower glucose clearance rate or diabetic fetopathy.[11]

Present in approximately 4% of all pregnancies, GDM is common enough that every obstetric practitioner needs to be aware of the disease and develop a consistent approach to its management. Because of its link to obesity, GDM may become a more common problem as the incidence of obesity increases. The prevalence of obesity (body mass index ≥30 kg/m2) in the US increased from 12.0% to 19.8% between 1991 and 2000.[12,13] Between 1991 and 1998, the increased prevalence of obesity included a 47% rise for women and a 70% rise for all individuals aged 18–29 years. This increased prevalence of obesity has theoretical implications for the rates of both GDM and type 2 diabetes (which eventually develops in at least half of women with GDM).

After accepting GDM as a legitimate disease entity with clinically significant morbidities, practitioners should evaluate the spectrum of recognized diagnostic and treatment strategies. Unfortunately, neither consistent scientific data nor a consensus of opinion exist with regard to diagnosis and treatment of GDM.

1. Diagnostic Strategies for Gestational Diabetes Mellitus (GDM)

The American College of Obstetricians and Gynecologists (ACOG) and the American Diabetes Association (ADA) agree that all pregnant women should have laboratory screening during the early third trimester unless no risk factors exist.[1,14] Such a risk-based screening approach would exempt only approximately 10% of a predominantly White population.[14] In the US, screening is typically done by measuring serum glucose 1 hour after the consumption of a 50g glucose beverage; a more specific diagnostic 3-hour 100g oral glucose challenge test is performed on women whose 50g glucose challenge exceeds 140 mg/dL (7.7 mmol/L).[14] Alternatives to that screening approach include the use of a random fasting plasma glucose measurement as the initial screening test,[15] a fixed dose of simple sugar (such as jelly beans)[16] or a single-step 75 or 100g oral glucose challenge test.[1] Universal screening is recommended by Diabetes UK; this organization advises a random plasma glucose measurement at the booking visit, at 28 weeks gestation, and whenever 1+ (≥250 mg/dL) glycosuria is detected at routine prenatal visits.[17] Diabetes UK recommends that plasma glucose >127 mg/dL (7.0 mmol/L) within 2 hours of a meal or >110 mg/dL (6.1 mmol/L) while fasting or >2 hours after eating should prompt a diagnostic 75g oral glucose tolerance test. A single report has shown that it may be possible to avoid glucose challenge testing in parous women with negative GDM screening during a prior pregnancy within 4 years; the risk for recurrent GDM in those circumstances is <1%.[18]

Thus, several practical options for screening exist. The choice of technique is not as important as the decision to screen, since the asymptomatic hyperglycemia of GDM will not be apparent without screening.

A final diagnostic debate regards the appropriate diagnostic test to be employed when the screening test is abnormal. The World Health Organization recommends a diagnostic 75g oral glucose tolerance test, considered positive if the venous plasma glucose is ≥127 mg/dL while fasting or ≥155 mg/dL 2 hours after the glucose challenge.[17] For decades, the 3-hour 100g oral glucose tolerance test has been utilized in the US, using thresholds defined by the National Diabetes Data Group (NDDG).[19] GDM has been diagnosed for women having two or more plasma glucose values greater than or equal to the threshold values of 105, 190, 165, and 145 mg/dL at fasting and hours 1 through 3 after the 100g oral glucose challenge. Since 2000, the ADA has recommended using a more conservative modification of the original data set (as modified by Carpenter and Coustan[20]), with threshold values of 95, 180, 155, and 140 mg/dL.[1] Without an evidence base to determine which diagnostic criteria are superior, the ACOG recognizes either standard as acceptable for diagnosing GDM.[14] Women diagnosed with GDM using the more conservative ADA criteria (rather than NDDG criteria) clearly have relatively small degrees of carbohydrate intolerance, and therefore might be expected to have a less extreme range of clinical manifestations. Shortly after the current ADA recommendations were published, a retrospective evaluation compared the obstetric outcomes for three cohorts of women: euglycemic controls, women with GDM (by NDDG criteria) managed with diet and insulin if needed, and women with intermediate carbohydrate intolerance (abnormal only by ADA criteria) who received no management for diabetes.[21] While the women with intermediate carbohydrate intolerance had a higher risk of neonatal hypoglycemia, they had no greater risk of macrosomia, cesarean delivery, or preeclampsia than the euglycemic control patients.

There are no uniform recommendations for the frequency of serum glucose monitoring or an ideal ‘target range’ for glycemic control. Women monitored frequently can expect to have more instances of hyperglycemia detected. In a study comparing home plasma glucose monitoring and weekly plasma glucose monitoring at office visits, the more frequently identified hyperglycemia in women with home monitoring led to a higher likelihood of insulin usage and a lower incidence of macrosomia, but no other differences in outcome.[22]

Recognition that the primary manifestation of relative hypoinsulinemia in women with GDM is postprandial hyperglycemia has prompted a comparison of the benefits of postprandial versus preprandial blood glucose monitoring.[23] Women randomized to postprandial blood glucose monitoring had babies with a lower mean birth weight (3.469 ± 0.668 vs 3.848 ± 0.434kg), and offspring were less likely to be macrosomic or have hypoglycemia. For both groups, insulin was adjusted to target ‘normal’ blood glucose results. Additionally, women monitored after meals lowered their glycosylated hemoglobin (HbA1c) to a greater degree and were less likely to undergo a cesarean delivery for cephalopelvic disproportion. The benefit of relatively strict postprandial glycemic control has also been demonstrated indirectly. In a small study of women with GDM randomized to receive only porcine regular insulin before each meal or a morning dose of Neutral Protamine Hagedorn (NPH) insulin, those receiving premeal short-acting insulin had babies with a smaller birth weight (3.079 ± 0.722 vs 3.943 ± 0.492kg) and a lower rate of macrosomia.[24] All patients performed home glucose monitoring at least 2 days per week. Insulin dosages were adjusted for patients receiving NPH insulin if two preprandial blood glucose values exceeded 100 mg/dL, and for those receiving only regular insulin if one 2-hour postprandial value exceeded 142 mg/dL. The maximum daily insulin dose was similar for both groups.

2. Treatment Strategies for GDM

Once diagnosed, a stepwise approach to GDM treatment is initiated. Instruction in proper diet and exercise is the foundation for treatment (see figure 1). If these strategies alone are unsuccessful, then pharmacotherapy typically is initiated.[1,14] Insulin remains the standard medication to treat diabetes in pregnancy, but many other possible treatment strategies are being evaluated. The immediate goal for treatment is to approach normoglycemia.

Fig. 1
figure 1

Stepwise approach to management of gestational diabetes mellitus (GDM). Solid arrow represents standard treatment strategies. Dotted arrow signifies nontraditional and investigational treatment options for GDM.

Full size image

2.1 Medical Nutrition Therapy

No published data exist to determine the optimal caloric content of a diet for women with GDM. The ACOG acknowledges 30 kcal/kg/day as a reasonable prescription.[14] While complex carbohydrates are consistently recommended over simple sugars, the optimal percentage of carbohydrate calories within the diet remains unresolved. Restricting carbohydrate content to 35–40% of calories decreases maternal postprandial glucose levels.[25] In a prospective study of women with GDM assigned to a low (<42%) or high (≥42%) carbohydrate diet, the cohort receiving a low carbohydrate diet were less likely to require insulin or to have a baby weighing >4kg.[25] Although diet is considered an important first step in management, and diet control can decrease the incidence of neonatal birth weight >90 percentile, it has no other proven benefit, including no demonstrated reduction in the likelihood of cesarean delivery.[26] A comprehensive review of the limited evidence base for this treatment strategy, as well as a plea for quality randomized control trials, has recently been published by Dornhorst and Frost.[27]

2.2 Exercise

Moderate exercise is considered an integral component in general diabetes management, and is considered safe for most women during pregnancy. Exercise leads to a short-term reduction in blood glucose levels. In a study of 12 women with GDM, 30 minutes of low and moderate intensity cycling exercise (35% and 55% of estimated maximum oxygen consumption, respectively) reduced blood glucose in a progressive and graded fashion.[28] After 45 minutes of exercise, there was no difference between the blood glucose levels of exercising and resting participants. In this small study, no impact of exercise on insulin levels was detected. No significant long-term benefits have been demonstrated for exercise in the management of GDM. Specifically, women with GDM who exercise have no reduction in neonatal birth weight, mean blood glucose levels, or HbA1c values.[29]

2.3 Insulin

If medical nutrition therapy does not provide adequate glycemic control, human insulin therapy is usually initiated. A typical approach in the US, supported by both the ADA and the ACOG, is to begin insulin therapy for GDM if medical nutrition therapy does not maintain fasting plasma glucose ≤95–105 mg/dL and/or 2-hour postprandial values ≤120–130 mg/dL.[1,14] No ideal insulin dosage or regimen has been identified for the management of GDM. Many empiric regimens are utilized, ranging from a single daily injection of intermediate-acting insulin (such as NPH insulin) to multiple daily doses of short-acting (such as regular human insulin) and intermediate-acting insulin. The ADA simply recommends that “self monitoring of blood glucose should guide doses and timing of the insulin regimen”.[1]

Prospective studies of insulin treatment for GDM generally demonstrate a reduction in the likelihood of macrosomia.[30,31] However, there are no other clear benefits from insulin treatment. Specifically, insulin treatment has not been proven to reduce the risk of cesarean delivery or birth trauma.[32]

Good, but not extreme, glycemic control has the most beneficial impact on fetal weight. In a prospective study of 334 women with GDM undergoing insulin therapy, those with a mean plasma glucose level of 87–104 mg/dL had an incidence of large for gestational age (LGA) and small for gestational age (SGA) neonatal birth weights similar to nondiabetic control women. As expected, poorer control, with mean plasma glucose concentration >104 mg/dL, was associated with a greater incidence of LGA neonatal birth weight. At the other end of the spectrum, extreme glycemic control (mean plasma glucose <87 mg/dL) was associated with a doubling of the risk (20%) for SGA birth weight.[30]

As potential alternatives to subcutaneous injection, several methods of insulin delivery across intact body surfaces have been attempted.[33] Disadvantages of subcutaneous injection include the cost and discomfort of injections, local degradation and suboptimal bioavailability, variability in absorption and relatively slow absorption, and the subsequent nonphysiologic peripheral concentration of insulin (greater than portal concentration). None of the alternative delivery systems have yet been successfully implemented clinically. Transdermal application has been unsuccessful as a result of inefficient transfer across an intact skin barrier. Oral insulin administration is not currently feasible due to proteolysis and the lack of a specific insulin transport mechanism across the intestinal epithelium. Inefficient transfer and low bioavailability have also so far prevented success for buccal or nasal administration of insulin. Although perhaps the most promising, pulmonary delivery of insulin is also complicated by inefficient delivery and limited bioavailability. No published data are available regarding the use of any of these insulin delivery systems in pregnant patients.

2.4 Insulin Analogs

The ideal insulin regimen for the management of GDM would eliminate the typical acute postprandial hyperglycemia and also prevent fasting hyperglycemia. Although a regimen of morning and evening mixed dosages of regular and NPH insulin can provide relatively higher levels of insulin after meals and overnight, the peak effect of both regular and NPH insulin occurs over a longer period of time than desirable to mimic the normal endogenous postprandial insulin surges (see table I). Therefore, exogenous administration of standard short- and intermediate-acting insulin can be complicated by initial postprandial hyperglycemia followed by hypoglycemia before the next meal.

Table I
figure Tab1

Pharmacokinetics of insulin and insulin analog formulations in the general patient population

Full size image

While preserving the receptor binding regions to maintain primary metabolic effects, other portions of the molecular structure of insulin can be modified to alter its pharmacokinetic properties, thus potentially creating a more physiologic effect. Structural modifications that inhibit self-association into dimers and hexamers lead to more rapid absorption after subcutaneous injection. Two short-acting human insulin analogs are currently commercially available. Both insulin lispro [Lys(B28)Pro(B29)] and insulin aspart [Asp(B29)] include modifications at amino acid sites of the B chain which are not critical in binding to the insulin receptor.

A single long-acting insulin analog is now commercially available. Human insulin is modified by extending the C-terminal end of the B chain with two arginine residues and substituting glycine for arginine at amino acid position 21 on the A chain to create insulin glargine. These modifications shift the isoelectric point from pH 5.4 to 6.7, making insulin glargine less soluble at a physiologic pH. Insulin glargine precipitates in the subcutaneous tissue following injection, delaying absorption and prolonging its duration of action.

When evaluated by HbA1c and mean blood glucose levels, insulin lispro and insulin glargine are at least as effective as regular and NPH insulin in individuals with type 1[34–38] and type 2 diabetes.[39] In addition, both insulin lispro and insulin glargine consistently reduce the incidence of hypoglycemic episodes in the same patient populations.[35–40]

The rapid onset and short duration of action of insulin lispro theoretically compensate for the postprandial endogenous insulin response deficits of women with GDM more effectively than regular human insulin. If insulin lispro could truly improve postprandial blood glucose levels, then it might further contribute to the favorable impact upon pregnancy outcomes, which are seen with tighter glycemic control following meals.[23,24] So far, a limited number of studies have been conducted using insulin lispro in pregnancy. Insulin lispro appears to have a similar impact on management of GDM as it does for type 1 and type 2 diabetes. Among women with GDM randomized to receive either regular human insulin or insulin lispro before each meal (both in addition to basal NPH insulin), those receiving insulin lispro had fewer hypoglycemic episodes before breakfast,[41] but no difference in overall blood glucose or HbA1c.[41,42] Furthermore, there was no difference in outcome between the management groups regarding neonatal birth weight or the incidence of macrosomia, cesarean delivery,[41,42] neonatal hypoglycemia or anti-insulin antibody levels.[41] A nonrandomized comparison of women with GDM receiving NPH insulin and either regular human insulin or insulin lispro also showed no difference in neonatal birth weight, cesarean delivery rate, or incidence of neonatal hypoglycemia. In that study, the cohort of 75 women receiving insulin lispro as their short-acting insulin ended pregnancy with a lower HbA1c value (5.8 vs 6.1%). For unstated reasons, insulin lispro was favored by the majority (19/22) of women followed in serial pregnancies who had been treated with both regular human insulin and insulin lispro.[43]

The clinical safety of insulin lispro has been demonstrated in several studies of nonpregnant patients.[44] In vitro studies also support the safety of insulin analogs. The mitogenic potential of insulin analogs correlates with their relative affinity for the insulin-like growth factor (IGF)-1 receptor and with occupancy time at the insulin receptor.[45,46] Because the mitogenic response to insulin requires a continuous stimulation by insulin for over 14.5 hours,[46] short-acting insulins (such as insulin lispro and insulin aspart) would not be expected to exert a mitogenic influence. Both of these short-acting insulin analogs, with insulin receptor affinities less than that of human insulin, also have mitogenic potencies less than human insulin.[45] The long-acting insulin glargine has an IGF-1 receptor affinity and mitogenic potency approximately 7 times greater than human regular insulin; the clinical implications of these findings remain to be determined.[45]

Data regarding the potential safety of insulin analogs in pregnancy are limited. A retrospective study of women receiving insulin lispro (n = 33) or regular human insulin (n = 27) during pregnancy showed no significant difference in the rate of fetal structural malformations.[47] The limitations of such a small retrospective study obviously preclude a confident declaration of fetal safety for insulin lispro. The majority of clinical studies addressing the safety of insulin analogs in pregnant women have explored a potential relationship between insulin lispro and the development of proliferative retinopathy. In one retrospective report of pregnant women with type 1 or 2 diabetes treated with insulin lispro beginning before or early in pregnancy, 3 of 10 women progressed from negative to proliferative diabetic retinopathy.[48] Two subsequent larger studies found no development of retinopathy associated with the use of insulin lispro during pregnancy in women with type 1 diabetes.[49,50] Many pregnancy factors have been identified in association with progressive diabetic retinopathy, leaving the true etiology unidentified. Although these conflicting reports cannot resolve whether the use of insulin lispro during pregnancy might be associated with the development of diabetic retinopathy, no significant risk in association with treatment for GDM would seem likely. The three women reported to develop retinopathy received insulin lispro for 20–23 weeks, far longer than would be appropriate for GDM.[48]

In the absence of adequate humans studies to evaluate the safety of insulin analogs for use during pregnancy, the US FDA pregnancy category assignment is based on the results of animal studies.[51] Reproduction studies performed in rats and rabbits have shown no harmful effects for insulin lispro, yielding an FDA pregnancy category B. With animal studies demonstrating no differences compared with regular human insulin, insulin aspart has been assigned an FDA category C. Insulin glargine has been assigned category C because five fetuses from two litters of rabbits exposed to the drug developed cerebral ventriculomegaly. Since insulin essentially does not cross the placenta and is used only during the latter half of pregnancy, teratogenicity is not anticipated due to insulin analog use in women with GDM. When insulin is needed in the management of GDM, the ADA recommends human insulin, noting “the use of insulin analogs has not been adequately tested in GDM”.[1] The ACOG acknowledge that insulin lispro “may be helpful in improving postprandial glucose concentrations” (see table II).[14] Additional clinical experience should define the proper role for these medications in pregnancy.

Table II
figure Tab2

Summary of gestational diabetes mellitus management recommendations from the American Diabetes Association (ADA) and American College of Obstetricians and Gynecologists (ACOG)

Full size image

2.5 Oral Antihyperglycemic Agents

An expanding spectrum of oral antihyperglycemic agents, with a variety of mechanisms of action, is being utilized to improve glycemic control in patients with type 2 diabetes. Although these agents have typically been avoided during pregnancy, a few reports have evaluated the use of oral antihyperglycemic drugs among pregnant women. Some,[53] but not all[52] studies reporting on the use of ‘first generation’ sulfonylurea agents during pregnancy demonstrated an association with prolonged neonatal hypoglycemia. Indeed, adult hypoglycemia is a common adverse effect of sulfonylurea agents, likely caused by their influence as a pancreatic β-cell secretogogue.

A potential teratogenic risk for first trimester chlorpropamide use has also been suggested. In an observational study of 20 women receiving oral antihyperglycemic agents during early pregnancy, half had babies with birth defects.[54] The majority of the defects were minor; four of six of those affected had external ear malformations and had been exposed to chlorpropamide. Such studies have influenced recommendations against the use of antihyperglycemic agents during pregnancy.[1]

Unlike first generation sulfonylurea agents which readily cross the placenta,[55] the demonstration that glyburide (glibenclamide) minimally crosses the human placental barrierr[55,56] paved the way for an evaluation of glyburide as a treatment for GDM. In a randomized prospective trial of women with GDM with fasting hyperglycemia, glyburide yielded outcomes similar to standard insulin treatment. Specifically, those women receiving glyburide had no difference in daily blood glucose concentration, perinatal mortality incidence or incidence of offspring large for gestational age (≥90 percentile), with a macrosomic (≥4kg) birth weight, congenital anomalies or hypoglycemia.[57]

Metformin, an insulin sensitizer, reduces hepatic glucose production in the presence of insulin and promotes peripheral glucose uptake. Metformin does not promote placental glucose uptake or transport, as demonstrated with an in vitro human placental perfusion model.[58] Although metformin is expected to cross the placenta based on its relatively small molecular weight,[55] the fact that it does not stimulate insulin secretion makes it a potentially attractive drug from the fetal perspective. Available internationally for decades, metformin was not released in the US until 1995. Metformin is often used to treat insulin resistance in nonpregnant women with polycystic ovary syndrome. A recent observational report documents the experience of 28 nondiabetic women with polycystic ovary syndrome who continued throughout pregnancy the metformin therapy initiated before conception.[59] Compared with a retrospective cohort of women with polycystic ovary syndrome who did not receive metformin during pregnancy, those who received metformin had a greatly reduced incidence of GDM (odds ratio 0.115; 95% CI: 0.014–0.938). No fetal malformations or neonatal hypoglycemia occurred among the offspring of those women receiving metformin throughout pregnancy. These reassuring results are contrasted by a Danish study which reports a 25-year experience treating 50 obese women with diabetes (gestational or type 2). Compared to nonobese, nonrandomized diabetic cohorts being treated with tolbutamide or insulin, those receiving metformin had a higher rate of preeclampsia (32%) and fetal death (8%). Beyond the potential bias created in this retrospective study by the nonrandomized treatment assignment based on weight, the impact on GDM may be obscured by the fact that one-third of the patients had type 2 diabetes.[52]

The α-glucosidase inhibitors, including acarbose and miglitol, limit intestinal carbohydrate absorption by competitive inhibition of an intestinal enzyme, which degrades complex carbohydrates. The subsequent reduction in postprandial glucose excursions directly targets the primary metabolic aberration in GDM. A Mexican study of six women with GDM treated with acarbose before each meal demonstrated a normalization of fasting and postprandial serum glucose levels.[60] As in nonpregnant patients, these women experienced persistent intestinal discomfort while taking acarbose. In nonpregnant patients, gastrointestinal adverse effects frequently lead to discontinuation of α-glucosidase inhibitors, and these agents are not as effective in lowering serum glucose levels as other oral antihyperglycemic medications.

No published human reproductive studies regarding the use of other oral antihyperglycemic agents, including thiazolidinediones (rosiglitazone or pioglitazone) and nonsulfonylurea secretogogues (repaglinide or nateglinide) have been identified.

In the absence of adequate human studies to direct recommendations for use during pregnancy, the FDA has assigned each oral antihyperglycemic agent to a pregnancy category based on animal studies (which include drug exposure in early pregnancy). Despite studies showing minimal transfer across the human placenta,[56] glyburide is FDA pregnancy category C, based on animal studies showing delayed maturation of fetal long bones. Based on a lack of any teratogenic effects in animal models, metformin and acarbose have been assigned as FDA category B. Repaglinide, nateglinide, rosiglitazone, and pioglitazone are all considered FDA category C. Fetal rats and rabbits exposed to rosiglitazone, pioglitazone, and repaglinide have an increased rate of growth restriction and fetal death.

Given that all potential oral antihyperglycemic treatments for GDM would be utilized only during the third trimester, teratogenicity is not a likely concern. Therefore, neonatal outcomes should be the most important factor in determining the safety of oral antihyperglycemic agents in women with GDM, and only limited neonatal outcome data are available. Only glyburide has been studied in a randomized prospective fashion, with no apparent fetal or neonatal harm among 201 exposed pregnancies.[57] This limited published experience precludes current advocacy for the clinical application of oral antihyperglycemic agents in the management of GDM.

3. Conclusion

Effective treatment for GDM must target the greatest morbidity: fetal macrosomia. Fetal size correlates with maternal serum glucose levels, making glycemic control a priority. Sharing the pathophysiology of insulin resistance, the treatment of GDM has traditionally included strategies that are beneficial in pregnant women with type 2 diabetes. Beyond the treatment foundation of medical nutrition therapy, exercise, and standard insulin regimens, the array of pharmacotherapy for type 2 diabetes has greatly expanded in the last several years. Slowed by potentially legitimate concerns about teratogenicity and fetal metabolic effects, limited data have accrued regarding the use of insulin analogs and oral antihyperglycemic agents during pregnancy. Among these new options, only insulin lispro and glyburide have been tested prospectively in randomized trials of women with GDM. In single studies of lispro insulin and glyburide, both of these agents compared favorably to standard insulin regimens, and neither appeared to cause any fetal or neonatal harm.[39,52] While acknowledged by the ADA and the ACOG, these seminal studies have not yet prompted a recommendation from either of those organizations to utilize insulin analogs or oral antihyperglycemic agents in the treatment of gestational diabetes (see table II). Although lacking an evidence base for many therapeutic strategies for GDM, the current ADA and ACOG guidelines provide a reasonable set of treatment recommendations. As obstetricians follow the lead of clinicians and researchers in the field of type 2 diabetes, we can look forward to future research replacing some of these empiric recommendations with evidence-based guidelines.