Insulin Analogs and Pregnancy

Pregnancy complicated by diabetes occurs in ∼ 4% of pregnancies in the United States.¹ With appropriate institution of intensive diabetes therapy and maintenance of glucose levels to achieve a lowering of hemoglobin A_1c (A1C) levels before and during pregnancy complicated by diabetes, the rate of fetal and maternal complications can be reduced to the rate observed among nondiabetic pregnancies.² Whether the pregnancy is classified as pregestational diabetes (occurring in women who have been diagnosed with type 1 or type 2 diabetes before pregnancy) or as gestational diabetes mellitus (GDM, occurring when a nondiabetic woman develops diabetes only during pregnancy), the goal of treatment is to maintain maternal glucose levels as near to normal as possible throughout the pregnancy.

Pregestational diabetes is a major risk factor for spontaneous abortions and congenital malformations, but the risk can be significantly reduced when hyperglycemia is controlled before conception and during the early first trimester when fetal organogenesis occurs.^3,4 In both pregestational diabetes and GDM, achieving near-normal glycemic levels throughout pregnancy can also decrease the prevalence of neonatal hypoglycemia, macrosomia, intrauterine death, and cesarean delivery.^5–7 Achieving near-normal glycemia means normalizing not only fasting glucose, but also postprandial hyperglycemia to reduce the risk of an adverse pregnancy outcome.^8–10 The infant mortality rate is nearly zero when blood glucose concentration is maintained at near-normoglycemia.¹¹ In both type 1 and type 2 diabetes, controlling pre- and postprandial hyperglycemia can best be accomplished during pregnancy by administering insulin via continuous subcutaneous insulin infusion from an insulin pump or by self-administration of multiple daily injections of an intermediate-acting basal human insulin (neutral protamine Hagedorn [NPH] insulin) combined with a short-acting or rapid-acting insulin. In GDM, achieving normoglycemia can initially be attempted with medical nutrition therapy and lifestyle modifications, but frequently, NPH insulin combined with a short-acting or rapid-acting insulin must be added to attain optimal glucose levels.

Throughout pregnancy, the placenta not only produces hormones that alter maternal carbohydrate and lipid metabolism, but it also controls the transplacental passage of glucose, fat, and protein to provide fuel energy and nutrients to the developing fetus. Maternal insulin is not transferred across the placenta unless the insulin is bound to IgG antibodies.^12,13 A study by Balsells et al.¹⁴ evaluated 50 women with GDM before and after treatment with human insulin, which is immunogenic, and found that 44% developed insulin antibodies. They also found that cord blood titers at delivery reflected the maternal antibody levels, indicating transplacental passage of human insulin when bound with maternal antibodies. In the same study, the status of the maternal insulin antibody did not appear to have an impact on fetal outcome.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Development of fetal hyperinsulinemia. Adapted from Ref. 15.

Because of the facilitated diffusion of glucose across the placenta, the maternal glucose level determines the glucose level of the fetus. When the maternal glucose level is high, the fetus becomes hyperglycemic, resulting in stimulation of the fetal pancreas to produce high levels of insulin (Figure 1).¹⁵

Because insulin is a growth factor, the high fetal insulin level increases intrauterine fetal growth, producing a newborn that is large for gestational age (LGA; weight > 90th percentile) or has macrosomia (weight > 4,000 g).¹⁶ Both conditions are frequently associated with birth trauma, especially shoulder dystocia, which occurs in 0.6–1.4% of fetuses weighing 2,500–4,000 g and in 5–9% of fetuses weighing > 4,000 g.¹⁷ Fetal hyperinsulinemia can also produce neonatal hypoglycemia after the infant is delivered and is no longer exposed to maternal hyperglycemia. Treatment of diabetic pregnant women with insulin to maintain maternal glucose levels near normal will prevent the development of fetal hyperinsulinemia in utero and decrease the risk of adverse fetal or neonatal outcomes associated with excess fetal growth.

The insulin requirement necessary to achieve near-normal maternal glycemia is based on the patient's weight and the number of weeks of gestation (Table 1).¹⁸

In recent years, molecularly engineered insulin analogs (the “new” insulins) have been developed and become available for the treatment of patients with hyperglycemia and diabetes.

The currently available rapid-acting insulin analogs are lispro, aspart, and glulisine; the long-acting insulin analogs are glargine and detemir. Lispro, aspart, and glulisine are similar rapid-acting formulations used primarily for mealtime insulin requirements, whereas glargine and detemir are long-acting basal formulations.^19–21 Also available are two biphasic formulations: biphasic aspart, which is a mixture of 30% aspart with 70% protaminated aspart, and biphasic lispro either consisting of 25% lispro and 75% protaminated lispro or 50% lispro and 50% protaminated lispro.

The insulin analogs are produced by recombinant DNA methods. Lispro, prepared by E. coli, differs from human insulin by two amino acid substitutions on the insulin β-chain; a lysine is substituted at position 28 and a proline at position 29, both of which are the reverse of human insulin. Aspart uses saccharomyces cerevisiae (yeast) to replace the proline at position 28 with an aspartic acid. Glulisine also has a double substitution, replacing asparagine with lysine at position 3 and lysine with glutamic acid at position 29. These modifications of the insulin β-chain inhibit the self-aggregation of insulin to form dimers and hexamers, resulting in monomers that are rapidly absorbed when administered subcutaneously and thus allowing a faster onset of action.

Glargine was the first long-acting analog to become commercially available. It has a glycine substituted for asparagine at position 21 on the insulin α-chain and two arginines added to the carboxyl terminal of the β-chain at positions 31 and 32. These amino acid modifications shift the isoelectric point of the insulin molecule from a pH of 5.4 toward a neutral pH, which delays its absorption and prolongs the duration of action.

Detemir is produced by the elimination of threonine at position 30 and the addition of myristic acid, a 14-carbon fatty acid chain, to the lysine located at position 29 of the β-chain. This acylation results in increased self-aggregation and greater reversible albumin binding that prolongs absorption to reach a maximum concentration in 6–8 hours and prolongs the duration of action up to 24 hours.

Rapid-Acting Insulin Analogs

Compared with regular insulin, the rapid-acting insulin analogs have a more rapid onset of action as well as a more rapid decline in concentration, which is advantageous for reducing postprandial hyperglycemia while avoiding late hypoglycemic events (Table 2).

Lispro, with an onset of action within 10–15 minutes, reaches peak concentration within 30–60 minutes and has a duration of action up to 3–4 hours.²⁰ Aspart is similar to lispro but may have a slightly longer time to peak concentration (40–50 minutes) as well as a slightly longer duration of action (3–5 hours).²² Glulisine also has a similar onset and duration of action, but has a slightly longer time to reach peak concentration (55 minutes).²³ Overall, when administered subcutaneously, the rapid-acting insulin analogs appear to have very similar pharmacodynamic and pharmacokinetic profiles.^24–26 Compared with regular human insulin, which has an onset of action of 30–45 minutes and a relatively prolonged effect of 2–3 hours, the rapid-acting insulin analogs have each demonstrated faster onset of action, earlier peak concentration, and briefer duration of action to achieve a more physiological dosing of insulin for lowering of prandial hyperglycemia while avoiding late-onset hypoglycemia.²⁷

Long-Acting Insulin Analogs

Glargine has a longer time to onset of action (1.5 hours) compared to NPH (0.8 hour) and ultralente insulin (1 hour), whereas the duration of action of glargine is longer (20.5 hours) than that of NPH (13.2 hours) and is similar to that of ultralente (19 hours) (Table 2).²⁸ Ultralente insulin is no longer available. In addition, several studies have reported less nocturnal hypoglycemia in both type 1 and type 2 diabetes, and in type 2 diabetes, there has been less weight gain with glargine than with NPH.^29–31

Detemir appears to have a similar profile to glargine, with slow absorption of 3–4 hours and a prolonged effect up to 24 hours^32,33 (Table 2); however, the best profiles are achieved when the insulin is administered twice daily. In clinical studies, detemir also demonstrated lower risk of nocturnal hypoglycemia and less weight gain compared to NPH in type 1 diabetes.^33–35

The long-acting insulin analogs, when used in nonpregnant patients with diabetes, appear to be most useful as basal insulin because of their long duration and lack of a peak.

Efficacy and Safety of Insulin Analogs During Pregnancy

All of the insulin analogs have been assigned a “safety during pregnancy” risk factor by the U.S. Food and Drug Administration (Table 3).³⁶ Lispro and aspart are the only insulin analogs currently classified as pregnancy risk Category B, which is the same risk category as regular insulin.

The benefit of controlling maternal glycemia during pregnancy by reducing hyperglycemia and avoiding hypoglycemia has been well documented.^2,3,37 Until the introduction of the insulin analogs, glycemic control during pregnancy was usually managed with diet and human insulin.

Lispro has been the most well studied during pregnancy. In pregnant women with type 1 diabetes, lispro has been shown to reduce A1C levels and postprandial glucose levels to lower or similar levels as those achieved with regular insulin but with fewer severe hypoglycemic events than with regular insulin.^38–43 Similar findings have been reported for aspart in the treatment of GDM when compared to regular insulin.^44,45 From these data, lispro and aspart appear to be as efficacious as regular insulin for improving glucose control while avoiding late hypoglycemic events during management of pregnancy complicated by diabetes. At present, there are no studies evaluating the use of glulisine during pregnancy.

There have been no comparator-controlled studies using the long-acting insulin analogs during pregnancy; however, there have been several case reports in which glargine was used for the treatment of hyperglycemia during pregnancy. These reports suggest that, compared to NPH, glargine results in improved glycemic control without any nocturnal hypoglycemia.^46–48

Although the insulin analogs appear to be effective and efficacious for controlling maternal hyperglycemia, there have been several concerns regarding their safety during pregnancy.

Safety of Insulin Analogs

Retinopathy and analogs. During the early weeks of pregnancy, women with the greatest decrease in A1C have an increased risk for progression of retinopathy. Reports early in the clinical use of lispro proposed that this insulin analog had a potential risk for developing or exacerbating diabetic retinopathy during pregnancy. These reports were based on the greater homology of lispro with insulin-like growth factor 1 (IGF-1), a growth hormone that has been indirectly implicated in the development of retinopathy among insulin-treated patients with type 1 diabetes.^49,50 Because lispro binds more than insulin to the IGF-1 receptor, it was postulated in an initial report in 1999 that if a patient were treated with lispro during pregnancy, there was a potential risk for increasing the development of diabetic retinopathy.⁴⁹ The authors had reported on three pregnant women treated with lispro who progressed from no background retinopathy to bilateral proliferative diabetic retinopathy.⁵¹ However, subsequent case reports comparing lispro to regular insulin have clearly shown that the incidence for progression of retinopathy is no different with lispro than with regular insulin.^52,53 Table 4 summarizes the results of studies evaluating retinopathy compared to regular insulin in pregnant women with type 1 diabetes, type 2 diabetes, or GDM.

Congenital anomalies and analogs. The increased potential for congenital malformations resulting from lispro has also been considered. However, when outcomes of lispro treatment during pregnancy were compared to those of pregnant women treated with regular insulin, there was no increased incidence of congenital anomalies among neonates whose mothers had been treated with lispro during gestation (Table 5).^39,41,42

Macrosomia and analogs. Placental transfer of insulin can occur when insulin becomes bound with immunoglobulin,^13,57 thereby contributing to excess fetal growth and development of macrosomia. In addition, lispro has not been detected in cord blood at delivery.⁵⁸ A study by Jovanovic et al.³⁸ compared 19 women with GDM treated with lispro with 22 women with GDM treated with human regular insulin. Although anti-insulin antibodies were similar in the two maternal groups, lispro was nondetectable in the cord blood, suggesting that it does not cross the placenta. In addition, no fetal or neonatal abnormalities were identified in either group.

The studies in Table 6 have evaluated the incidence of neonatal macrosomia among women treated with lispro compared with human regular insulin.

There are no controlled human studies of glargine use during pregnancy. The effect of glargine in pregnancy has only been evaluated in animal studies in which toxicological effects were seen at doses that were seven times the highest human dose of both glargine and NPH insulin.⁶¹ These effects resulted from the hypoglycemia that both insulins induced. There appeared to be no effect on reproduction or embryo-fetal development among the studies conducted in rats.⁶¹ In addition, it is not known whether glargine is excreted in breast milk. However, because human insulin is excreted in human milk, caution has been recommended for nursing mothers who are using glargine.⁶²

There have been many case reports in which pregnant women with diabetes were treated with glargine. Table 7 summarizes the outcomes of those reports. Although glargine may improve control compared to NPH, it is not possible to determine the frequency of macrosomia from these reports.

There have been no clinical trials in human pregnancy treated with detemir, the other long-acting insulin analog, although clinical trials are in the early stages. Animal studies for embryotoxicity and teratogenicity have not demonstrated any difference between detemir and NPH insulin. Although it is clear that human insulin and all of the rapid-acting insulin analogs appear in milk directly proportional to the serum levels of insulin or insulin analog achieved in the maternal bloodstream, it is currently unknown whether detemir is excreted in breast milk. Therefore, women who use detemir must also be cautioned when lactating.⁶⁷ However, since insulin is not absorbed through the gut, even if the insulin appears in human milk, it is not biologically active. Although glargine and detemir reportedly have been used during pregnancy, there is insufficient data to support their use as a safe basal insulin during pregnancy. Additional studies are needed to evaluate the efficacy and safety of glargine and detemir compared with the current use of NPH basal insulin during pregnancy complicated by diabetes.

Conclusions

For pregnant women with diabetes, the timely initiation of insulin treatment to maintain preprandial glucose levels < 90 mg/dl and postprandial glucose levels < 120 mg/dl has reduced the risks of fetal and maternal complications. When compared with human regular insulin, the rapid-acting insulin analogs have been shown to be efficacious in reducing hyperglycemia during pregnancy, with a safety profile that resulted in a lower incidence of neonatal complications. NPH insulin is the only basal insulin that has been adequately studied in pregnancy. The long-acting insulin analogs do not yet have sufficient safety evaluation in clinical studies to warrant their use during pregnancy.

In addition to diet and lifestyle modifications, the treatment algorithm in Table 1 is the currently recommended guideline for all insulin-requiring pregnant women with type 2 diabetes or GDM or pregnant women with type 1 diabetes who require multiple daily insulin injections and are not using an insulin pump. Early and intensive treatment intervention will improve the outcome of all pregnancies complicated by diabetes.