Due to the increased use of newer antiepileptic drugs (AEDs), we agree with Hernandez-Diaz et al. that it is vital to assess the efficacy and safety of the same drugs in the general population and in pregnant women, especially regarding teratogenicity.¹ We also agree that valproic acid use is not acceptable during pregnancy.

The authors excluded spontaneous abortion cases, which reveals a bias because even “safer” drugs like lamotrigine may play a role in these cases. Fetotoxicity of the less “teratogenic” drugs was not considered because only stillbirths and live births were included.¹ Lamotrigine levels drop by more than 50% in pregnancy.² The actual serum levels of lamotrigine are unclear in this study. If the doses had been escalated to target nonpregnant levels, there may have been more teratogenic effects.

Finally, the most efficacious drug—valproic acid—was the most teratogenic. Valproic acid has prevented maternal seizures in a significant number of cases and more than any other newer drugs mentioned in this study. Uncontrolled seizures are embryopathic. Stillbirths, microcephaly, mental retardation, and nonfebrile seizure disorders occurred more frequently in the offspring of women with seizure disorders.³

This finding in itself may negate the teratogenic effect directly attributable to valproic acid. Valproic acid should not just be viewed in terms of its teratogenicity because up to a low therapeutic dose it is still safe (figure 1¹).

Author Response

We excluded spontaneous abortions because there was not enough information as to whether the embryo had—or was going to develop—a malformation. We would have included spontaneous abortions if we had postmortem reports, including the results of chromosome analyses, since 50% of spontaneous abortions are associated with chromosome abnormalities. There is no current information on any of the spontaneous abortions that have occurred among the women enrolled in the North American AED Pregnancy Registry. In general, pregnancy registries are not a good study design to observe the occurrence of spontaneous abortions, as so many occur before the typical gestational age when women enroll. However, spontaneous abortions were excluded from all exposure groups, not only from the lamotrigine group. The proportion of women excluded because of spontaneous abortions, withdrawals, or losses to follow-up did not vary significantly with the specific anticonvulsant. Therefore, it is unlikely for this selection bias to explain the lower risks found for lamotrigine.

We took into account dose changes during the first trimester and did not find a dose effect for lamotrigine.^4,–,8 Maheshwari et al. question whether serum levels would be a better exposure measurement. We are obtaining these on a subset of women whose doctors' records include levels. We will be analyzing the associations of outcomes, like malformations, in the exposed fetuses in those subsets when the sample size is large enough to be informative.

We did not find a positive association between seizures during pregnancy and a higher risk of malformations. Others found that neither the number of seizures during pregnancy nor having epilepsy increase the risk of having children with major malformations. Therefore, we found it unlikely that epilepsy itself could explain the risks found for valproic acid.

In our population, even low valproate doses were associated with a twofold increased risk of malformations. However, we agree that there are risks associated with seizures during pregnancy and that most pregnant women need to maintain their medication. The challenge is to assess the safest drug for each individual patient from those that are effective for her type of epilepsy.