CLINICAL OBSTETRICS AND GYNECOLOGY Volume 50, Number 4, 959971 r 2007, Lippincott Williams & Wilkins

From Educated Guess to Accepted Practice: The Use of Oral Antidiabetic Agents in Pregnancy
ODED LANGER, MD, PhD Department of Obstetrics and Gynecology, St Lukes-Roosevelt Hospital Center, Columbia University, New York, New York

Abstract: When pharmacologic therapy is required, oral antidiabetic agents have been almost universally endorsed as first line drugs in the treatment of gestational diabetes mellitus (GDM) based on welldesigned studies that have found no association between these agents and congenital malformations. These agents have an efficacy comparable with insulin in their ability to facilitate achievement of targeted levels of glycemic control on all GDM severity levels and in obese patients. Therefore, level of glycemic control achieved, not the mode of therapy is the key to improving outcomes in GDM. Key words: gestational diabetes, glyburide, level of glycemic control, perinatal outcome

Nearly 3% to 10% of the 4.5 million pregnancies annually in the United States are affected by gestational diabetes mellitus (GDM). Pharmacologic therapy will be required in 20% to 60% of these pregnancies to maintain targeted levels of glycemic control. Insulin has traditionally been considered the gold standard for the management of GDM based on its effectiveness (50% to 80%) in achieving established levels of glucose control. Antidiabetic oral agents would have to provide comparable results to be considered efficacious alternatives. Oral antidiabetic agents are commonly prescribed for nonpregnant type 2 diabetic patients. The main objection to their use in pregnancy is the risk for the development of congenital anomalies,1 fetal compromise, and fetal hypoglycemic episodes through direct stimulation of
Correspondence: Oded Langer, MD, PhD, Department of Obstetrics and Gynecology, St. Lukes-Roosevelt Hospital Center, 1000 Tenth Avenue, New York, NY 10019. E-mail: odlanger@chpnet.org
CLINICAL OBSTETRICS AND GYNECOLOGY /

the fetal pancreas.2 The historic ban on the use of oral antidiabetic agents in pregnancy has been based on scant evidence of case reports and one study in particular on fetal anomalies in 50 poorly controlled diabetic women before pregnancy.1 In 2000, the results of a randomized study3 comparing glyburide and insulin use in pregnancy demonstrated similar outcomes and the ability to achieve established levels of glucose with either drug. The results of this study helped to swing the pendulum toward acceptance of glyburide in the treatment of GDM. Treatment approval, reflected in editorials, clinical opinions and follow-up studies with comparable results, was followed by endorsements for its use by the North American Diabetes in Pregnancy Study Group (2002)4 and in 2005 by the Fifth International Workshop on Gestational Diabetes (sponsored by the American Diabetes Association).510 Thus today, the use of oral antidiabetic drugs, especially glyburide, has become the standard of care in the management of gestational diabetes in many centers and in private practices throughout the United States. Of note, oral antidiabetic agents were used in the United Kingdom and South Africa in the early 1960s but with a disregard for rigid evidence-based standards (grossly underpowered, flaws in design and conduct, and first generation antidiabetic drug use rarely prescribed today). Ironically today, in light of the overwhelming evidence that has been collected from basic science studies addressing pharmacologic characteristics of these drugs in randomized and case-controlled studies, practitioners in Europe show a reluctance to use the drugs!
VOLUME 50 / NUMBER 4 / DECEMBER 2007

959

960

Langer
glucose intolerance and approximately 60% loss when clinical type 2 diabetes develops. Hepatic insulin resistance and relative insulin deficiency also leads to an increase in hepatic gluconeogenesis, which further worsens the degree of hyperglycemia. Thus, prediabetes may include patients with metabolic syndrome, gestational diabetes, and impaired glucose tolerance and fasting plasma glucose (FPG). All represent stations on the road to type 2 diabetes. It is well recognized that obesity and pregnancy are factors that affect increased insulin resistance. The additional stimulation on the bcells alone is another reason for b-cell deterioration. GDM women, either lean or obese, are characterized by further insulin resistance and an inability to increase insulin secretion as a compensatory response. Therefore, type 2 and gestational diabetes are actually the same disease with different names that occur at different stages on the glucose continuum of natural deterioration of glucose intolerance.13 It should be noted that type 2 diabetes is defined on the basis of a selected threshold on the glucose intolerance continuum. These thresholds are fluid, for example, in the past, the threshold required for a diagnosis of type 2 diabetes was fasting of 140 mg/dL, whereas currently it is 126 mg/dL. The thresholds were created on the basis of the rate of complications associated with a particular threshold. New research may reveal the need for different thresholds. In summary, GDM represents an early stage of the deterioration continuum toward type 2 diabetes. The foundation for this conceptual framework is based on the similarities in risk factors and metabolic and endocrine abnormalities: type 2 diabetes and GDM are heterogeneous disorders whose pathophysiology is characterized by peripheral insulin resistance, impaired regulation of hepatic glucose production, and declining b-cell function. Among different ethnic groups, both forms have been diagnosed in varying prevalence. The b-cell function in 4209 newly diagnosed type 2 diabetic women was only 51% as shown in the United Kingdom Perspective Diabetes Study (UKPDS) followed by a progressive decline to approximately 28% after 6 years.14,15 The b-cell loss may begin more than a decade before the diagnosis and is, therefore, the consequence of a disease process that has remained undetected. Thirteen to 14 years after diagnosis, b-cell function and insulin secretion may plunge to zero.16 Therefore, if oral antidiabetic agents are the gold standard for therapy in type 2 diabetes, from

This review provides the rationale for the use of oral antidiabetic agents in pregnancy as an efficacious alternative to insulin. It describes how research from the basic sciences (placental transfer) to clinical studies (perinatal outcome) have lead to significant evidence on which to base management recommendations for the use of these drugs in the treatment of GDM and type 2 diabetes. The association between level of glycemic control, treatment modalities, disease severity, and obesity are evaluated.

The Foundation for the Use of Oral Antidiabetic Agents in the Management of GDM
Normal pregnancy is generally characterized by peripheral insulin resistance, which causes an increase in b-cell activity (insulin secretion). The result will be an approximate 50% decrease in insulin-mediated glucose disposal and a 200% to 250% increase in insulin secretion. The basic metabolic defect in women with GDM suggests the limited capacity of pancreatic b-cells to increase insulin secretion to compensate for the enhanced insulin resistance of pregnancy (impairment in the first phase of insulin secretion).11 In addition to their diminished b-cell reserve, the increased insulin resistance found during pregnancy has been shown to be present before and after pregnancy.11 Insulin resistance, that is, impaired tissue response to insulin, occurs early in the disease process leading to type 2 diabetes.12 Insulin resistance alone, however, does not cause diabetes. Most obese people do not develop type 2 diabetes despite increased insulin resistance; there needs to be an accompanying relative insulin deficiency. Initially, early in the deterioration stages (prediabetes, GDM), the pancreatic b-cells respond by secreting more insulin with the increase in insulin resistance to maintain normal glucose tolerance (compensatory hyperinsulinemia). With the increase in insulin resistance, there is a decline in insulin secretion with an accompanying glucose intolerance characterized by postprandial and/or fasting glucose elevation. Thus, during the prediabetes stage, glucose intolerance (GDM) and metabolic syndrome can be identified years before the development of the threshold that defines type 2 diabetes. As b-cell function further declines, hyperglycemia becomes more severe. Forty percent of b-cell mass may be lost in individuals who have

From Educated Guess to Accepted Practice

monotherapy to combination therapy in relation to the disease severity, it is reasonable to speculate that women with GDM will respond to oral therapy with even greater success. These pharmacologic agents should be prescribed because of ease of use, noninvasive strategy, and cost-effectiveness as an alternative therapy option offered to patients. However, the above can be provided as a management alternative only after there is conclusive proof of the efficacy and safety of these drugs in comparison with other therapies.

961

lurealike drugs could be used in the treatment of type 2 diabetes. Obviously, this is not the case.

Will the Use of Sulfonyureas in Pregnancy, Especially Glyburide, Cause an Earlier Exhaustion of b-cells in GDM Patients in Later Life?
Approximately 70% of women who experience GDM will develop type 2 diabetes within 10 years after delivery. The risk to develop type 2 diabetes varies markedly in different studies and in patients of different ethnic and geographic backgrounds.17 The UKPDS demonstrated that insulin deficiency was a progressive condition that did not appear to be affected if a patient received sulfonylurea or metformin therapy.14,15 The pharmacologic effect of antidiabetic agents on plasma glucose will gradually disappear after the drug use is discontinued, that is, wash-out period. In diabetes prevention trials with patients with impaired glucose tolerance, the effect of metformin, acarbose, and troglitazone began to disappear shortly after the drugs were discontinued.18 In addition, the use of troglitazone in the Prevention of Diabetes study of women with previous GDM, reported improvement in insulin secretion after the wash-out period. This may be the result of the selection of younger women in comparison with the other studies. Similar effects were seen in our experience with GDM patients treated with insulin therapy. When the drug was discontinued by the patient, they remained within the targeted levels of glucose control for 1 to 2 weeks before the wash-out effect set in. In summary, it is not reasonable to assume that within the 8 to 12-week window of GDM therapy with glyburide or any other oral antidiabetic agents that the drug will either contribute to an escalation or deterioration of pancreatic function. In fact, if this was the case, no oral therapy with sulfonylureas or sulfony-

Different oral antidiabetic agents act upon diverse mechanisms of action to correct or improve the pathologic lesion responsible for glucose intolerance. Therefore, these drugs provide an enhanced approach to the treatment of type 2 and gestational diabetes (Fig. 1). Furthermore, combination therapies will further improve the effect of these drugs on glucose metabolism. Insulin, in contrast to the oral agents, is designed to mimic the physiologic secretion of endogenous insulin. The basal insulin is supplied by the administration of intermediate or long acting at bedtime or before breakfast and at bedtime. The meal-related (glucose excursion) insulin includes the use of short acting insulin. Sulfonylureas that require functional pancreatic b-cell for their hypoglycemic effects have been used to treat type 2 diabetes for many decades. They bind to specific receptors (SUR1: sulfonylureas, repaglinide, nateglinide) on b-cell plasma membrane, resulting in closure of potassium ATP channels with a parallel opening of calcium channels. The resulting increase in cytoplasmic calcium stimulates insulin release. The principal significance of these drugs is to improve insulin secretion. Enhanced insulin secretion suppresses the production of hepatic glucose, the main contributor to fasting hyperglycemia. It lowers glucose toxicity and facilitates insulin secretion after meals thereby reducing postprandial hyperglycemia. These drugs as a secondary effect can also improve peripheral tissue sensitivity to insulin as shown by several studies.1921 The peak plasma level of glyburide (known as glibenclamide in Europe) when given as a single agent occurs within 4 hours and its absorption is not affected by food intake. Glyburide is extensively metabolized in the liver and its metabolites are extracted in bile and urine to equal extent. Six metabolites of glyburide are formed by the human and baboon placentas and livers but the rates of their formation are different between organs and species.22 The half-life of glyburide is approximately 10 hours and results in elimination and a steady-state time of (1/2 life-time 5) 50 hours. This component needs to be addressed because there is frequent dose adjustment (in all

Oral Agents: Pharmacologic Characteristics (Tables 1 and 2)

962

Langer
Selective Characteristics of Oral Antidiabetic Agents
Duration of Antidiabetic Action (h) Decrease in FPG (mg/dL) 60-70 24-72 24 12-24 24 Unknown 6-12 >24 >24 Unknown Unknown 35-40 20-30 0.7-1 0.7-1 60-70 60-70 1.5-2 1.5-2 6.2 3-7 3-4 2 URI nasopharyngitis, headache GI symptoms Weight gain, fluid retention GI symptoms 60-70 1.5-2 4 Hypoglycemia, weight gain Decrease in Hemoglobin A1C (%) 1.5-2 Half-life Time (h) 10

TABLE 1.

Class Sulfonylureas Chlorpropamide Glipizide Glyburide (DiaBeta, Micronase) Glyburide (Glynase) Secretagogues Repaglinide Nateglinide Biguanides Metformin Thiazolidinediones Pioglitazone Rosiglitazone Glucosidase inhibitors Carbose Miglitol DPP-4 inhibitors sitagliptin

Side Effects Hypoglycemia, weight gain

FPG indicates fasting plasma glucose; GI, gastrointestinal; URI, upper respiratory infection.

oral agents). The starting dose is 2.5 mg orally in the morning. If the targeted level of glycemia is not attained, add 2.5 mg to the morning dose. If indicated (after 3 to 7 d), add 5 mg in the evening. Thereafter, increase the dose by 5-mg increments, up to a total of 20 mg/d. If the patient does not achieve targeted levels of glycemic control, add long-acting insulin to the regimen or assign the patient to insulin therapy alone (Fig. 2). Hypoglycemia is the main side effect of glyburide therapy and is dose-related. It may occur in 11% to 38% of nonpregnant type 2 diabetics and is a greater risk for the older patient. These rates are significantly lower in pregnancy because prospective mothers are generally younger and healthier. In our original randomized study, we evaluated the number of hypoglycemic episodes throughout pregnancy. We found a significant decrease in the glyburide compared with the insulin-treated patients.3 In another study, using continuous blood glucose monitoring for 3 days, we reconfirmed our original findings; however, the testing time was limited.23 Furthermore, when insulin and glyburide were compared for weight gain in pregnancy,

no difference was noted between the groups. Although glyburide has been found to be associated with increased weight gain in nonpregnant type 2 (long-term) diabetes, GDM is of relatively short-term duration (weeks).3

Biguanides
Metformin (Glucophage) affects the glycemic profile by improving insulin sensitivity at the cellular level (insulin sensitizer). The cellular effects of metformin are to counteract insulin resistance and basal plasma insulin levels. Its chief glucose-lowering action suppresses hepatic glucose output by reducing the excessive rates of glucose production, which in turn decreases gluconeogenesis. The drug also decreases hepatic glycogenolysis and increases insulin-stimulated glucose uptake in skeletal muscles. This action causes a reduction in the recognized toxic metabolic effects of hyperglycemia (glucose toxicity) and fatty acids (lipotoxicity) in type 2 diabetes. Its metabolic effects include reduction of fatty acid oxidation; increase in splanchnic glucose turnover; improvement of lipid profile with the decline of triglyceridemia, fatty acids, and

From Educated Guess to Accepted Practice

TABLE 2. Oral Antidiabetic Drugs Pharmacologic Effects
Pregnancy Cross Category Placenta Excreted in Breast Milk

963

Drug Sulfonylureas Glimepiride (Amaryl) Glipizide (Glucotrol) GlipizideGITS (Glucotrol XL) Glyburide (DiaBeta, Glynase, Micronase) Meglitinides Nateglinide (Starlix) Repaglinide (Prandin) Biguanide Metformin (Glucophage) Glitazones

Mechanism of Action Increase insulin secretion

Daily Maintenance Dosage

Unknown Unknown 1-8 mg/d as a single daily dose with breakfast or the first main meal of the day Minimal Unknown 5-40 mg/d as a single daily dose 30 min before breakfast 5-20 mg/d as a single dose with breakfast

No

No

0.75-12 mg/d as a single daily dose with breakfast or the first main meal or as 2 divided doses

Increase insulin secretion C C Decreases hepatic Gluconeogenesis, increase B insulin sensitivity Increased insulin sensitivity, decrease hepatic glucose production C C Unknown Unknown Unknown Unknown 1-16 mg/d in 2 to 4 divided doses within 15-30 min before meals Yes No 1500-2550 mg/d in 2 or 3 divided doses with meals

Pioglitazone (Actos) Rosiglitazone (Avandia) a-Glucosidase inhibitors Acarbose (Precose) Miglitol (Glyset)

Unknown Animals Unknown Animals

15-45 mg/d as a single daily dose (with or without meals) 4-8 mg/d as a single daily dose or in 2 divided doses in the morning and evening (with or without meals)

Slow absorption of carbohydrates in the intestine B B Unknown Animals Unknown Animals 150-300 mg/d in 3 divided doses at the start of each day 150-300 mg/d in 3 divided doses at the start of each meal

low-density lipoprotein cholesterol while to some extent increasing high-density lipoprotein cholesterol and decreasing intestinal absorption of glucose and utilization. Metformin, too, stabilizes weight reduction but does not stimulate insulin secretion. Therefore, it does not cause hypoglycemia in either diabetic or control subjects or stimulate the fetal pancreas to over secrete insulin. Metformin was introduced after the withdrawal of phenformin (a first generation biguanide). Lactic acidosis with the use of metformin is infrequent (0.03 cases/1000 patient-years), one tenth that of phenformin. Lactic acidosis in-

creases with renal dysfunction and patient age. Its peak plasma level, given as a single agent, occurs within 4 hours. Although food intake reduces absorption, it should not be administered without it, because it can produce gastrointestinal intolerance. Metformin is a highly water-soluble drug that is not metabolized and is eliminated unchanged through the kidneys by tubular secretion; glyburide is metabolized in the liver. Peak plasma concentrations are shortlived; in patients with normal renal function the plasma half-life (t1/2) for metformin is 2 to 5 hours with almost 90% of an absorbed dosage eliminated within 12 hours.24 Renal clearance of

964

Langer
Insulin Insulin secretion secretion
Sulfonylureas (+) Megalitinides & analogs (+) (repaglinide) Exenatide (byetta) (+)

muscle muscle uptake of glucose uptake of glucose

Thiazolidinediones (+) Biguanides (metformin)(+)

Hyperglycemia Hyperglycemia
Thiazolidinediones (-) Biguanides (metformin) (-) Glyburide (-)

hepatic hepatic glucose production glucose production

FIGURE 1.

Oral antidiabetic agent: mechanism of action.

metformin occurs more often by tubular secretion than glomerular filtration with minimal binding of metformin to plasma proteins. On the other hand, glyburide is cleared by the liver and kidneys with 99.8% bound to plasma proteins. As a result, with the use of metformin in pregnancy, its therapeutic level should be

adjusted with the method of clearance because of the increased glomerular filtration rate in pregnancy. Metformin should be introduced gradually in 500-mg or 850-mg increments to a maximum effective dose of 2000 mg/daily. The absolute maximum dose is 2550 or 3000 mg/d. Metformin is contraindicated in the presence of

Gestational or Type 2 Diabetes

1. 2. 3. 4. 5.

Dose and Adjustment 2.5 mg, morning Increase by 2.5 mg first week Add evening 5 mg Increase AM 5 mg Increase PM 5mg Maximum: 20 mg/d

Postprandial blood glucose Levels of 120 to 180 mg/dL (6.7 to 9.4 mmol/L) Or Fasting blood glucose of 95 to 140 mg/dL (5.3 to 7.8 mmol/L)

Glyburide

Combination Therapy

Glyburide + Metformin Glyburide & Acrabose Glyburide + Insulin (pm)

Insulin

FIGURE 2.

Algorithm of drug administration and decision-making.

From Educated Guess to Accepted Practice

renal disease. Clinical trials in 2 centers examined the effect of metformin as single therapy in nonpregnant obese subjects with type 2 diabetes and in poorly controlled sulfonylurea-treated subjects. In both studies, the mean decline in plasma glucose concentration was about 60 mg/ dL and there were observed beneficial effects on plasma lipid levels.15,25

965

Solving the Placental Transfer Enigma of Oral Antidiabetic Agents

Fetal anomalies, macrosomia, and neonatal hypoglycemia have all been attributed to the use of these drugs without inclusion of data explaining failure to achieve established levels of glycemic control. Moreover, these assumptions have not been confirmed by specific correlation with the degree of maternal-fetal placental transfer of these drugs. There have been 3 main concerns against the use of oral hypoglycemic agents in pregnancy on the basis of clinical observations mostly in the 1960 to 1970s, and 1980s primarily in type 2 diabetes using first generation sulfonylurea drugs: (1) increased rate of congenital anomalies,1 (2) the potential for fetal macrosomia caused by direct stimulation of the fetal pancreas resulting in hyperinsulinemia, and (3) the increased rate of hypoglycemia owing to fetal hyperinsulinemia. These potential negative effects were cited in a single case report of 3 infants whose mothers received chlorpropamide and another mother of an infant given acetohexamide, with another case report describing prolonged symptomatic neonatal hypoglycemia.2 The recirculating single-cotyledon human placental model is widely used to characterize the transport and metabolism of numerous drugs and nutrients. It has been recognized as a safe in vitro surrogate for human placental transfer. It is a practical model because it facilitates the study of intact human placenta independent of fetal metabolism. Finally, each experiment can be validated with the addition of antipyrine as a reference point for the level of transfer. With this in mind, we sought to evaluate if sulfonylurea drugs readily cross the placenta. We used a recirculating single-cotyledon placenta model in vitro to characterize the maternal-tofetal term placentas perfused immediately after delivery. As glyburide has a molecular weight of 494 units, it is one of the largest oral hypoglycemic agents and also the most commonly used second-generation sulfonylurea. Peak serum levels after a single oral dose of 5 mg occur at 4 hours and range from 112 to 360 ng/mL. Over 99.8% of glyburide is extensively bounded to albumin and metabolism occurs through hydroxylation in the liver to inactive metabolism of glyburide across the human placenta. We demonstrated2830 that second-generation oral hypoglycemic agents, especially glyburide, do not significantly cross the diabetic or nondiabetic placenta. Fetal concentrations reached no more than 1% to 2% of maternal concentrations. These findings were supported by our clinical

Thiazolidinediones
Thiazolidinediones are category C drugs that should only be prescribed in pregnancy if the likely benefit justifies the potential risk to the fetus. These drugs should be used cautiously with confirmation of liver disease or abnormal liver testing (alanine aminotransferase levels >2.5 times the upper limit for the laboratory), and in patients compromised with heart disease or a history of heart failure. Liver function needs to be measured before initiation of therapy and bimonthly during the first year of treatment. There is evidence of substantial weight gain with the use of both drugs. There is scant evidence of these contraindications in pregnancy and, therefore, most patients should be able to use the drug.

a-Glucosidase Inhibitors
This group of drugs slows the absorption of carbohydrates in the intestines thereby reducing the postprandial rise in blood glucose.26 Because of gastrointestinal side effects with the use of these drugs, gradual dosage increments, over weeks to months is recommended after initiation of therapy. Acarbose (Precose), miglitol (Glyset), and voglibose, currently in clinical use, may be added to most other available therapies. The use of these drugs in pregnancy is limited (100 patients to date) despite the fact that the action of acarbose is within the gastrointestinal tract and is not transferred by the blood stream to the placenta. However, it is less effective than glyburide in decreasing glucose levels to those required in pregnancy. Recently, in a small-scale study of 59 patients of which 29 were randomized to acarbose therapy and 30 to placebo, the success rate with acarbose was 58% with a significantly lower weight gain when compared with the placebo group. The study was underpowered to evaluate any outcome variables.17,27 Its primary use in pregnancy should not be as monotherapy but rather in combination with glyburide and possibly metformin.

966

Langer
The majority of drugs used in pregnancy cross the placenta although very few will cause adverse fetal outcome. Company information provided by Glucophage, Bristol-Myers Squibb, 1997, explains that there is a partial placental barrier to metformin. Studies of transplacental transport of metformin reported that it noticeably crosses the placenta,3839 but has minimal affect on transplacental glucose flux.28 In addition, 2 studies measured metformin levels in umbilical cord blood at delivery. In the first study, arterial and venous umbilical samples were not analyzed separately and it is unclear whether the results originate from maternal or fetal samples.40 In the second study, cord blood samples for metformin were taken from 15 patients. The time between the last drug intake before sampling ranged from 4 to >48 hours. The authors concluded that metformin freely passes the placenta and that the fetus is exposed to therapeutic concentrations. However, there were no teratogenic effects identified in this small sample-size study.41 Finally, metformin transfer into human milk is minimal. The infant concentration during breast feeding is <0.4% of the maternal concentration, corrected for body weight. This is substantially lower than the arbitrary cutoff of 10% used to guide lactating mothers. Thus, this finding implies safety; metformin is not contraindicated in lactating patients.42,43 The results of reproduction studies in rats and rabbits remain controversial. One study demonstrated no teratogenicity at doses up to 600 mg/kg/d, approximately twice the maximum recommended human dose on the basis of mg/m2. In other studies, there has been evidence suggesting that metformin induces a low incidence of malformations in rats. In another study, 2-cell mouse embryos were exposed to different levels of metformin. Although lower levels comparable with plasma concentration during treatment did not affect the blastocyst development rate, the highest concentration resulted in a marked delay in mice development. It has been suggested in a single-cotyledon ex vivo human perfusion model using 10 term placentas of uncomplicated pregnancy that there is minimal transfer and fetal accumulation of rosiglitazone. The maternal concentration perfused in this study corresponded to plasma levels reported for an 8-mg oral dose.44 In contrast, another study using the same technique found that the transfer rate of rosiglitazone is 90% of antipyrine that is used as the reference point. Thus, rosiglitazone readily crosses the term placenta. In addition,

study3 in which glyburide was not detected in umbilical artery blood or showed increased fetal insulin levels. Furthermore, we sought to rule out the possibility that the time from administration of the drug to the mother may affect the findings in the fetus. In 13 mothers who received the drug shortly before delivery, glyburide dose was within therapeutic concentrations ranging from 50 to 150 ng/mL, whereas arterial cord blood levels of glyburide were undetectable with high-performance liquid chromatographic analysis. Koren31 in analyzing the results of our studies suggested a pharmaco-physiologic alternative explanation for glyburides lack of transferability across the placenta by using the classic factors affecting placental permeation. The author advocated for plasma protein binding as the explanation for glyburides lack of transferability across the placenta. Tolbutamide, chlorpropamine, and glipizide have high protein binding (>96), with glyburides protein binding of 99.8%. This results in a 4% circulating tolbutamide versus 0.2% circulating glyburide (20-fold more molecules available to cross the placenta). Another unique feature of glyburide is that the protein binding is stable at serum concentrations exceeding 10-fold levels encountered during clinical use. The short lifetime of glyburide (4 h) and its rapid clearance rate (1.3 0.5 mL/kg/min) also contribute to the lack of transferability. A potential problem is that albumin levels decrease physiologically in pregnancy owing to an increase in glomerular filtration rate. Thus, there is increased transfer of glyburide. However, when this hypothesis was tested, it was shown that decreased albumin concentration associated with pregnancy is unlikely to affect the deposition of glyburide.32 In addition, it was demonstrated in perfusion studies that glyburide is actively effluxed by a transporter other than P-glycoprotein. Alternatively, it is possible that a smaller portion of glyburide is carried by P-glycoprotein or, that most of the fetal load is pumped to the mother by an as yet unidentified placental transport system.33 Recently, Gedeon et al34 demonstrated that glyburide is transported by BCRP and MRP3. It was also found to be an inhibitor of BCRP, PGP, MRP1, MRP2, and MRP3. The elimination of the threat of glyburide crossing the placenta and the adverse affects to the fetus (malformations and hypoglycemia) enhances the potential for the use of glyburide as a vigorous alternative pharmacologic agent in the management of GDM patients.3537

From Educated Guess to Accepted Practice

placental tissue retention of rosiglitazone is low and the drug does not affect placental tissue viability and functional parameters.45 These observations need to be reconfirmed to eliminate the disparity in the findings and duplicated on the placentas of diabetic women. If the findings confirm that the drug does not cross the placenta, it may lay the groundwork for the introduction of a new drug with tremendous potential in the treatment of GDM, metabolic syndrome, and polycystic ovarian syndrome (PCO) women during pregnancy.

967

The Dogma of Oral Hypoglycemic Agents and Fetal Anomalies

In approaching the dilemma of fetal congenital anomalies and their association or lack thereof to a specific pharmacologic agent, 2 basic questions need to be addressed: (1) does the drug cross the placenta? and (2) if it does, is there an established association between the drug and fetal anomalies? It should be noted that the majority of drugs do cross the placenta but, in general, few have teratogenic effects on the fetus. The dogma that congenital anomalies are associated with the use of oral antidiabetic drugs was perpetuated with the results of a study, demonstrating an increased rate of congenital anomalies in 50 type 2 patients with hyperglycemia before conception (HbA1c >8%).1 However, it is difficult to determine if the rate of anomalies was caused by the use of the drug or the existing hyperglycemia. Notelowitz46 randomized study in 1971 of patients treated with first generation sulfonylureas and insulin resulted in only 2 births with anomalies. One infant treated with tolbutamide had choanal atresia and the other treated with insulin had Fallot syndrome. His conclusion was that sulfonylureas were safe for use in pregnancy.46 Three studies in the 1990s also found no association between oral antidiabetic agents and congenital malformations. Towner et al47 treated 332 type 2 patients with oral antidiabetic agents or insulin before pregnancy. The authors showed, by using a stepwise logistic regression, that mode of therapy did not have an adverse significant effect; however, level of glycemia and maternal age were significant factors contributing to the rate of anomalies. We48 found similar findings in a retrospective analysis of 347 type 2 diabetic women exposed to different oral antidiabetic agents, insulin, and diet therapy before and during the first trimester of pregnancy. Again, it was the blood glucose and not the mode of therapy that had the

net effect on the rate of anomalies. Finally, Gilbert et al49 performed a systematic review and metaanalysis of 8 studies from 1996 to 2004. The principal outcome measure was major malformations after metformin use for polycystic syndrome. Treatment with metformin in the first trimester was associated with a statistically significant 57% protection effect with an anomaly rate of 1.7% in the metformin group compared with 7.2% in a matched control group. The authors concluded that there is no evidence of an increased risk for major malformations with metformin use during the first trimester. Two other studies reported the use of metformin in women with PCO syndrome. Glueck et als50 study suggested that metformin is safe and useful in the reduction of GDM in women with PCO. They evaluated 33 nondiabetic women with PCO who were treated with metformin prospectively and 39 nondiabetic women not treated with metformin evaluated retrospectively. The development of GDM was 3% in the former and 23% in the latter. In the study by Jakubwicz et al,51 the authors sought to investigate the effect of preconceptual use of metformin on early pregnancy loss by evaluating 65 women with PCO receiving metformin compared with 31 women untreated with the drug. The early pregnancy loss in the former was 11% and 58% in the untreated group. All 62 infants in the metformin-treated group were normal with the exception of 1 infant who had chondrodysplasia. The data in both studies suggest that the use of metformin preconception and during the first trimester is not associated with major fetal malformations or fetal hypoglycemia after birth. To date, the cumulative available evidence suggests that the cause of anomalies in diabetes in pregnancy is the failure to achieve the established level of glycemic control and not the use of oral antidiabetic drugs, especially since replacement by the second generation of hypoglycemic drugs.

Should Type 2 and PCO Patients Treated With Metformin Preconception Remain on the Medication Throughout Pregnancy?
Although metformin crosses the placenta, it is a class B drug. It may be a potentially attractive alternative to insulin in the management of GDM patients because its pharmacologic action reduces hepatic glucose production and insulin resistance, However, until such time that a wellcontrolled study will be conducted in type 2 and

968

Langer
A major finding in the UKPDS was that type 2 diabetes is a progressive disease with increasingly greater b-cell deterioration that, over time, requires increased and/or multiple therapies to maintain near normal glycemic control. In the first 5 years of the study, 70% of the patients treated with glyburide achieved the desired goal. In years 3, 6, and 9, the desired goals were achieved by fewer than 55%, 40%, and 30% of patients, respectively, using a single agent (insulin, sulfonylurea, or metformin). The authors (DCCT and UKPDS) concluded that the improvement in glycemic control rather than a specific therapy was the primary factor responsible for reduction of risk factors. These subjects represent phases 2 and 3 of type 2 diabetes.16,25 Studies of pregnant diabetic women, including our prospective GDM studies,55 challenged health care professionals to implement standards for improved glycemic control to decrease disease complications. Despite the success of intensified therapy with insulin to control blood glucose levels, a less invasive, more patient friendly alternative that enhances patient compliance while achieving similar perinatal outcome has become a welcome alternative. As GDM is characterized by a milder glycemic profile and occurs 2 to 10 years earlier than type 2 diabetes, the use of oral antidiabetic agents with GDM patients should prove to be even more effective than its use with type 2 diabetes. Moreover, it will be reasonable to assume that the success rate of therapy for GDM should be at least 70% or higher than was achieved in the UKPDS. In our randomized study3 that compared glyburide and insulin therapies, the goals of treatment were to achieve mean blood glucose of 90 to 105 mg/dL, fasting blood glucose 60 to 90 mg/dL, preprandial concentration 80 to 95 mg/dL, and postprandial concentration <120 mg/dL. We demonstrated that glyburide use in GDM patients is as effective as insulin when 82% of the glyburide and 88% of the insulin patients achieved the targeted levels of control. There were 8 glyburide-treated women (4%) receiving the maximal dose who failed to achieve the established levels of glycemia; they were reassigned to insulin therapy. In the early 1970s, in another randomized study, it was shown that 80% of patients treated with first generation sulfonylurea: tolbutamide, chlorpropamine, and/or diet were able to maintain their blood glucose within the targeted levels of glycemia (<150 mg/dL). In contrast, only 38% of the insulin patients were able to achieve this level probably because of poor compliance.46

GDM patients, metformin use should not be endorsed. We are currently awaiting the results of an ongoing randomized trial adequately powered addressing the possible effect of metformin in pregnancy (MiG study-Metformin in Gestational Diabetes). Can targeted levels of glycemic control be achieved in type 2 diabetic pregnant patients? The targeted level of glycemic control in nonpregnant patients is <150 mg/dL to prevent macro and microcardiovascular complications. In pregnancy, the desired level of glycemic control to prevent perinatal complications such as fetal macrosomia, stillbirth, metabolic, respiratory, and hematologic complications is much lower. It is reported to be <100 mg/dL and/or <120 mg/dL postprandial determinations.53,54 Glyburide, metformin, and the majority of other oral antidiabetic drugs reduce FPG by 2 to 4 mmol/L accompanied by a decrease in HbA1c of 1% to 2% (a decrease of 1% = 20 to 30-mg glucose). Therefore, type 2 diabetic patients who commonly reach higher plasma glucose levels (HbA1c >10%) may not be able to achieve the targeted levels of control with oral antidiabetic agents in pregnancy. To date, there is paucity of data on the use of these drugs in pregnancy in comparison with insulin therapy. One can speculate that type 2 diabetic patients in phase 1 of the disease (early onset) may benefit from monotherapy or combination or oral antidiabetic drugs, that is, glyburide and metformin. More severely ill type 2 diabetic pregnant women (phases 2 and 3) will most likely fail to achieve desired levels of glycemic control with oral antidiabetic agents.

The Association Between Targeted Levels of Glycemic Control and Oral Antidiabetic Agents in the Management of GDM
The achievement of established levels of glycemic control in diabetic patients, both pregnant and nonpregnant, became axiomatic as a major factor for improving outcome. The mainstay of glycemic control in pregnancy has been the use of insulin therapy. The use of intensified therapy in diabetes management has become the standard of care to obtain the above goal. The UKPDS14,15,25 and the Diabetes Complications Clinical Trials (DCCT) strongly supported its use with type 1 and 2 diabetic patients to achieve and maintain near normal HbA1c and to deter diabetic complications.

From Educated Guess to Accepted Practice

The glycemic threshold assigned to a given study will determine the rate of success. In addition, the number of samples and time-related glucose observations, that is, preprandial or postprandial, may all mask and influence the relation between pregnancy outcome and level of glycemic control. In general, with reported results of high morbidity rates in macrosomia or large for gestational age (LGA), questions should be raised if the targeted levels of glucose have truly been achieved or properly defined.

969

Clinical Studies
Sulfonylureas (glyburide) are currently the only oral agent group studied extensively in GDM women that have provided conclusive evidence for the efficacious and safe treatment of GDM. Glyburide and metformin were found to be as effective as insulin in maintaining desired glycemic levels and result in comparable outcomes.3,52 However, other oral antidiabetic agents may in the future prove to have an even greater therapeutic effect in controlling the abnormal levels of glycemia. We3 conducted a randomized clinical trial with 404 women comparing glyburide and insulin-treated patients. The 2 groups did not differ significantly in the rates of preeclampsia, cesarean section, and level of glycemia prior and subsequent to treatment. Additionally, no significant difference was found in the overall rates of small-for-gestational age, macrosomia, Ponderal Index, and the rate of perinatal complications between the groups. There was no identifiable

trend for one of the groups when all were found to be nonsignificant. Furthermore, the 95% confidence interval for the difference of the mean was found to be relatively small, which suggests the unlikely possibility of beta errors. When patients were stratified by level of glycemic control to evaluate the impact of glycemia on the rate of abnormal fetal size, no difference was found between the 2 treatment groups but there was a significantly higher rate of large infants in the poor glycemic category (Fig. 3). The results obtained in our randomized study were similar to the results obtained in our quasirandomized intensified treatment study.55 Since our original publication in 2000, several investigators reported their clinical experience with glyburide.5665 Each demonstrated the effectiveness of glyburide therapy to achieve glycemic control. However, different studies used different criteria to define success rate. For example, Chmait et al60 in 46 patients evaluated failure of glyburide therapy. Failure was defined when the maximum glyburide dose could not maintain fasting plasma <110 mg/dL and 1hour postprandial <140 mg/dL. Approximately 81% of the patients achieved these goals. Jacobson et al64 with 122 women on insulin and 137 treated with glyburide instructed patients to test blood glucose 4/daily: fasting and either 1 or 2 hours postprandials (per individual provider preference). Targeted goals were fasting 100 mg/ dL, 1-hour 155 mg/dL, and 2-hour 130 mg/dL. Eighty-six percent of glyburide and 63% of insulin patients achieved these goals. However, the reported macrosomia rate was 25%.

A
20
PERCENT

Mean 96 + 8

B
20 15 10 5 0

Mean 120 + 8

15 10 5 0

Macrosomia

LGA

Macrosomia

LGA

Insulin

Glyburide

FIGURE 3. The association between level of glycemic control and rate of large infants by treatment modalities (A, good glycemic control and B, poor glycemic control). Modified from Semin Perinatol. 2002;26:215224.

970

Langer
and comparable groups (compliant vs. noncompliant subjects). Thus studies reporting similar success rates may result in significantly different perinatal outcomes. which in turn may lead to erroneous conclusions on cause and causation. What remains unanswered is: will glyburide be as effective as insulin at all severity levels of GDM? And, is there a dose limitation above which the efficacy of glyburide will decrease in comparison with insulin? We found that glyburide and insulin are equally efficacious for GDM treatment at all severity levels of diabetes when FPG on a glucose tolerance test was between 95 and 139 mg/dL. As the level of disease severity increases, the success rate for achieving established levels of glycemic control decreases. The majority of patients (71%) will require, on average, up to 10 mg daily dose of glyburide to achieve established levels of glycemic control. After stratifying patients by GDM severity, no significant difference was found in neonatal size, metabolic complications, and the composite outcome between the 2 treatment modalities. For adjustment of the potential confounding effects of several factors, we performed logistic regression analysis when the primary outcome was LGA. We found that the mean blood glucose [odds ratio (OR) 1.99, 95% confidence interval (CI) 1.04-3.83], severity of GDM (categorized by the fasting plasma from the oral glucose tolerance test) (OR 2.13, 95% CI 1.09-4.18), previous macrosomia (OR 3.73, 95% CI 1.70-8.81), and weight gain in pregnancy (OR 3.81, 95% CI 1.81-8.05) were the only significant contributors. Again, treatment modality, parity, and prepregnancy weight [body mass index (BMI)] were found to be noncontributors. Therefore, reaching established levels of glycemic control and not the mode of therapy is the key to improving pregnancy outcome in GDM women.68 Obesity, in and of itself, is a precursor of potential adverse outcome in pregnancy. Diet-treated GDM patients, overweight (BMI 26 to 29) or obese (BMI Z30), are associated with adverse pregnancy outcome regardless of the level of glycemic control. In contrast, for those treated with insulin therapy who achieve established glycemic levels pregnancy, outcome will be comparable with those of normal weight patients.69 As glyburide therapy has become an alternative to insulin in the treatment of GDM, we sought to evaluate if the success rate to achieve targeted levels of glycemic control and perinatal outcome were comparable with both modalities. Patients who achieved targeted levels of glycemic control had comparable perinatal outcome for both treatment modalities. Therefore, appropriate utilization of glyburide

Several studies have sought to identify the predictors of glyburide failure therapy. Rochon et al66 studied 101 GDM women requiring pharmacologic therapy while testing 4/daily. Criteria for success was achieving fasting between 60 and 90 mg/dL and 2-hour postprandial <120 mg/dL. Seventy-nine percent of the women achieved targeted levels of glycemic control. The authors concluded thaty predicting glyburide failure is difficult, but failure does not appear to be associated with increased adverse pregnancy outcomes. However, in this study, pregnancy outcome included shoulder dystocia 10%, macrosomia (in the success group) 16%, and cesarean delivery approximately 40%. These outcome rates are higher than expected in well-controlled diabetic women. Kahn et al63 analyzed 95 GDMs receiving glyburide therapy. The overall success rate was 81%. Criteria for failure were 20% of fasting blood glucose determinations at Z95 mg/dL and 1-hour postprandial Z140 mg/dL. Patients were instructed to take the glyburide 30 minutes before breakfast and dinner when the initial dose was individualized on the basis of patient weight and degree of hyperglycemia. This administration criterion is unconventional and deviates from standard recommendations in the literature. The perinatal outcome was associated with 27% LGA infants and 12% preeclampsia. Their conclusion was that glyburide was more likely to fail in women diagnosed in pregnancy of older age, multiparity, with higher fasting glucosey Is it possible that with the above perinatal outcomes, the majority of patients were undiagnosed type 2 diabetic women? Recently, we sought to identify predictors of treatment failure in GDM in 379 glyburidetreated women. Failure of glyburide therapy is largely dependent on the physicians ability to recognize and adequately adjust the glyburide dose. Physician intervention can preclude an avoidable failure. GDM severity, obesity, early gestational age at diagnosis, maternal age, and parity are all known factors that influence the success rate in the treatment of the diabetic patient independent of pharmacologic agent. All of these were found to be noncontributing variables to failure rate of glyburide therapy.67 Success in truly achieving level of glycemic control may vary from study-to-study because of different criteria for success, failure to administrate the maximal dose, different doses and administration algorithms, length of therapy, type of patient (severity, ethnicity, and obesity),

From Educated Guess to Accepted Practice

25 Diet 20 15 10 5 0
Composite LGA Macrosomia Metabolic C/S

971

Insulin

Glyburide

FIGURE 4. The relationship between treatment modalities (diet, insulin, and glyburide) and perinatal outcome in wellcontrolled obese patient.

therapy in obese patients will result in pregnancy outcome comparable with those treated with insulin (Fig. 4).70 In our experience, glyburide has become the drug of choice for use in GDM when pharmacologic intervention is required, regardless of GDM severity level and obesity. The noninvasive, cost-effective,71 patient-friendly regimen lends itself more readily to potential patient compliance. Although both treatment modalities show comparable perinatal outcome, it seems from our and other investigators experience that oral therapy is more readily accepted than insulin injections. However, failure to achieve established levels of glycemic control, regardless of the choice of treatment modality and physician failure to provide the appropriate drug algorithm and dose will result in adverse perinatal outcome. In the near future, we will require studies with appropriate sample sizes and power that will evaluate different oral antidiabetic drugs and combination therapies, that is, glyburide and acarbose, glyburide and metformin, and also an oral antidiabetic agent in combination with insulin to optimize patient/physician choices in the treatment of diabetes in pregnancy. Note: Following is an abbreviated reference list. The complete reference list is available on line at: www.clinicalobgyn.com

References
3. Langer O, Conway DL, Berkus MD, et al. A comparison glyburide and insulin in women with gestational diabetes mellitus. N Engl J Med. 2000;343:11341138.

7. Greene MF. Oral hypoglycemic drugs for gestational diabetes [editorial]. N Engl J Med. 2000;343:11781179. 9. Saade G. Gestational diabetes mellitus: a pill or a shot? [editorial]. Obstet Gynecol. 2005;105: 456457. 23. Yogev Y, Ben-Haroush A, Chen R, et al. Undiagnosed asymptomatic hypoglycemia: diet, insulin, and glyburide for gestational diabetic pregnancy. Obstet Gynecol. 2004;104:8893. 33. Kraemer J, Klein J, Lubetsky A, et al. Perfusion studies of glyburide transfer across the human placenta: implications for fetal safety. Am J Obstet Gynecol. 2006;195:270274. 50. Glueck CJ, Wang P, Kobayashi S, et al. Metformin therapy throughout pregnancy reduces the development of gestational diabetes in women with polycystic ovary syndrome. Fertil Steril. 2002;77:520525. 59. Kremer CJ, Duff P. Glyburide for the treatment of gestational diabetes. Am J Obstet Gynecol. 2004;190:14381439. 63. Kahn BF, Davies JK, Lynch AM, et al. Predictors of glyburide failure in the treatment of gestational diabetes. Obstet Gynecol. 2006;107: 13031309. 64. Jacobson GF, Ramos G, Ching J, et al. Comparison of glyburide and insulin for the management of gestational diabetes in a large managed care organization. Am J Obstet Gynecol. 2005;193:118124. 67. Langer O, Most O, Monga S. Glyburide: predictors of treatment failure in gestational diabetes [abstract]. Am J Obstet Gynecol. 2006;195:S136. 68. Langer O, Yogev Y, Xenakis E, et al. Insulin and glyburide therapy: dosage, severity level of gestational diabetes, and pregnancy outcome. Am J Obstet Gynecol. 2005;192:134139.