Children exposed to valproate in utero have significantly lower IQs at age 3 than do children exposed to other antiepileptics during gestation, according to findings from the interim analysis of a large international study.
The drug previously had been associated with a higher rate of birth defects in children exposed prenatally. The combination of findings strengthens a recommendation to avoid valproate as a first-line antiepileptic in women who may bear children, Dr. Kimford J. Meador said in an interview.
“Valproate poses a special risk for both congenital malformations and for cognitive impairment,” said Dr. Meador, principal investigator on the Neurodevelopmental Effects of Antiepileptics Drugs (NEAD) study. “Since there are other therapeutic options, it would seem prudent to try those first. At a minimum, it is critical that physicians inform women of this risk when prescribing valproate so that they may make an informed choice.”
NEAD is an ongoing study of 309 children, including three sets of twins, born in either the United States or the United Kingdom from 1999 to 2004, whose mothers were taking a single antiepileptic drug (AED): carbamazepine, lamotrigine, phenytoin, or valproate. The children are being followed to age 6. Dr. Meador, professor of neurology, Emory University, Atlanta, and his associates reported a planned 3-year interim analysis (N. Engl. J. Med. 2009;360:1597–605).
All of the 303 women in the study were taking the drugs for a seizure disorder. Their mean age at delivery was 30 years. Most women were well controlled on their AED, with about 80% having no seizures during their pregnancy.
Most children (258) underwent cognitive assessment at either 2 or 3 years of age, or at both ages. Of these, 73 (28%) had been exposed to carbamazepine, 84 (32%) to lamotrigine, 48 (19%) to phenytoin, and 53 (21%) to valproate.
IQ scores were adjusted for factors that could significantly affect cognitive development, some of which were maternal IQ, age at delivery, education, type of epilepsy, and seizure frequency.
Children exposed to valproate had the lowest mean IQs of any of the exposure groups (92)—significantly lower than those of any other treatment group. The mean IQ in those exposed to carbamazepine was 98; to lamotrigine, 101; and to phenytoin, 99. These did not vary significantly from one another.
The association of valproate with reduced IQ held after adjustment for confounders in both a linear regression and subgroup analysis, the investigators said.
They also examined whether the IQ scores were related to AED dosage. In this analysis, only valproate maintained a significant dose-response relationship. Additionally, higher maternal IQs were associated with higher child IQs in all of the treatment groups except valproate.
The results are consistent with several European studies that have found poor cognitive outcomes in children exposed to the drug prenatally, the investigators said. A Finnish study reported a mean reduction of 13 points in verbal IQ in valproate-exposed children, compared with controls. A British study found that valproate exposure increased developmental delays, increased special education needs, and reduced verbal IQ, compared with unexposed children and children exposed to carbamazepine and phenytoin.
The findings of both physical and cognitive problems with prenatal exposure show that the drug probably is not safe for use at any time during pregnancy, said Dr. Michael Privitera, director of the Cincinnati Epilepsy Center and another of the NEAD investigators.
“The neural tube defects [with which valproate is associated] occur during the first trimester, so there has been a question whether we might be able to use valproate later in pregnancy. This study shows that the answer is no, because cognitive development in the fetus occurs during the third trimester,” Dr. Privitera said in an interview.