More information is sorely needed about the possible effects of oral glucose-lowering agents during pregnancy. Although type 2 diabetes is a growing problem among women of childbearing age, there are virtually no prospective, randomized data about the safety of commonly used oral diabetes drugs on the developing fetus, said Dr. Montoro, professor of clinical medicine and ob.gyn. at the University of Southern California, Los Angeles.

Pregnant women have been routinely excluded from clinical trials, due largely to liability concerns of pharmaceutical firms that might fund such trials, he said.

Yet abundant data clearly show that uncontrolled maternal diabetes is teratogenic. It appears that in many cases, adverse fetal outcomes that have been attributed to oral glucose-lowering agents—including various anomalies, stillbirths, macrosomia, and neonatal hypoglycemia—were probably due to the diabetes itself, he said.

Insulin works, of course, but it's not usually an attractive option for women who are used to taking pills to control their diabetes.

Dr. Montoro offered his expert advice, based on mostly limited data from animal studies, anecdotal reports, and retrospective reviews. In the best scenario, the woman comes in for preconception counseling, allowing for optimization of glucose control—down to a hemoglobin A1c level of less than one standard deviation above the lab's normal mean—before pregnancy. Diet and exercise are standard therapy; insulin should be prescribed if glucose levels continue to be elevated.

If she has been taking oral agents, the dose should be adjusted to achieve optimal diabetes control while on adequate contraception, then switched to insulin once HbA1c is optimized and she's ready to become pregnant.

Unfortunately, the more common scenario is that the patient presents when she is already pregnant. Some women will have stopped taking the drugs on their own when they discovered they were pregnant. “It's very important to counsel these patients that the risk of anomalies and other complications is probably related to their diabetes, rather than the medications,” Dr. Montoro said.

But as a practical matter, none of these drugs are approved for use during pregnancy. The limited data that are available suggest that glyburide is probably safe throughout pregnancy, but all other agents should be switched to insulin. Metformin and thiazolidinediones (TZDs) are probably safe in early pregnancy, but data are insufficient to recommend their use throughout gestation.

Oral antidiabetes drugs got a bad name during pregnancy mainly from early studies on the first-generation sulfonylureas tolbutamide and chlorpropamide. A 1962 paper reported 14 perinatal deaths in offspring of 22 women taking chlorpropamide and 4 deaths in offspring of 17 women taking tolbutamide. There were no congenital anomalies or neonatal hypoglycemia. All the women were in poor glycemic control by today's standards, and the study was retrospective (Diabetes 11[suppl.]:98-101, 1962).

That study has been widely quoted, even though those two agents are rarely used any more, and there have been at least a dozen other studies published over the last three decades that have found no abnormalities associated with sulfonylureas, particularly the newer ones.

For example, a study that compared diet alone in 125 women during the first 8 weeks of pregnancy, oral agents (chlorpropamide, glyburide, or glipizide) in 147 women, and insulin in 60 women showed no significant difference in major or minor congenital anomalies (Diabetes Care 18[11]:1446-51, 1995).

In the only randomized trial of oral hypoglycemic agents during pregnancy, there were no differences in neonatal outcomes such as large for gestational age, hypoglycemia, anomaly, or stillbirth between 201 women randomized to receive glyburide after 8 weeks' gestation and 203 treated with insulin (N. Engl. J. Med. 343[16]:1134-38, 2000).

Even fewer data are available for the other classes of diabetes drugs. One retrospective review of 160 pregnancies found that 32% of women taking metformin during pregnancy had preeclampsia, compared with 7% of those on sulfonylureas and 10% of those on insulin. Stillbirths were also higher with metformin (Diabetic Medicine 17[7]:507-11, 2000).

TZD data are limited to case reports, and most are of troglitazone, which is no longer on the market. There is one case report on rosiglitazone and none on pioglitazone. No ill effects from TZDs have been reported in human pregnancies, but troglitazone was associated with fetal death and growth retardation during mid to late gestation in animals. There are no pregnancy data for the oral agents repaglinide or nateglinide, Dr. Montoro said.